<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[PassiveLogic - Medium]]></title>
        <description><![CDATA[The first fully autonomous platform for buildings. - Medium]]></description>
        <link>https://medium.com/passivelogic?source=rss----b3718cadbd32---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>PassiveLogic - Medium</title>
            <link>https://medium.com/passivelogic?source=rss----b3718cadbd32---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 16:52:07 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/passivelogic" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Benchmarking Differentiable Swift]]></title>
            <link>https://medium.com/passivelogic/benchmarking-differentiable-swift-270da84f4fe4?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/270da84f4fe4</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[pltech]]></category>
            <category><![CDATA[automatic-differentiation]]></category>
            <category><![CDATA[swift]]></category>
            <category><![CDATA[swift-programming]]></category>
            <dc:creator><![CDATA[Brad Larson]]></dc:creator>
            <pubDate>Mon, 26 Feb 2024 15:51:48 GMT</pubDate>
            <atom:updated>2024-02-26T15:51:48.243Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SZ7gZMQkys70r_y5LYvfXQ.jpeg" /></figure><p>PassiveLogic’s <a href="https://passivelogic.com/autonomous-buildings/">truly autonomous control of buildings</a> and more is enabled by the use of <a href="https://quantumalliance.org/">physics-based digital twin simulations</a>. To match these simulations to the real world and use them to make optimal control decisions, we need to be able to run these simulations faster than real-time. Additionally, we need to be able to optimize them at the edge, not on massive servers. <a href="https://medium.com/passivelogic/intro-to-differentiable-swift-part-2-differentiable-swift-25a99b97087f">Differentiable programming</a> and <a href="https://medium.com/passivelogic/intro-to-differentiable-swift-part-1-gradient-descent-181a06aaa596">gradient descent</a> optimization are the key items needed to make this possible.</p><p>As described in <a href="https://passivelogic.com/?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=launchevent#:~:text=Differentiable%20Swift">our recent Launch Event video</a>, we believe that the experimental language feature of differentiable Swift is uniquely capable of performing these simulations and identifying optimal control paths. What led us to that conclusion?</p><p><strong>The speed of differentiable Swift</strong></p><p>Swift is a memory-safe and strongly-typed systems language. These properties make it ideal for building robust and reliable control systems. However, what really sold us on Swift for our simulations was its <em>performance</em>.</p><p>Early in the life of the company, we created several benchmarks of representative building simulations and the equipment operating within them. In one such benchmark, gradient descent optimization of a building model in differentiable Swift was 10,000,000 times faster than an industry-standard building optimizer without gradient descent.</p><p>I want to highlight another benchmark, one that we just published to our <a href="https://github.com/PassiveLogic/differentiable-swift-examples">open source differentiable Swift examples repository</a>. This is the benchmark behind <a href="https://youtu.be/xr8sV3GdhVk?t=4265">our recent public speed claims</a>, which show a building simulation created in differentiable Swift operating <strong>238X faster than PyTorch</strong> and <strong>322X faster than TensorFlow</strong>. But…what is being measured here, and why is that important to us?</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fxr8sV3GdhVk%3Fstart%3D4265%26feature%3Doembed%26start%3D4265&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dxr8sV3GdhVk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fxr8sV3GdhVk%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/8afa6b23ab007711847aac2ebcb8c411/href">https://medium.com/media/8afa6b23ab007711847aac2ebcb8c411/href</a></iframe><p>Simulating heat transfer in a building is critical for control purposes. How the sun, air flow, and equipment heats or cools parts of the building are key variables needed to determine system adjustments. How do we maximize comfort while <em>also</em> minimizing energy use?</p><p>We can describe heat transfer in terms of relatively simple equations. However, we may not know some of the equations’ parameters, and therefore need to optimize those parameters to match the known behavior of the system. That known behavior could be temperature measurements from sensors as a function of time, the recorded external weather, or any other data we have about the building or equipment.</p><p>The system we’re modeling in this benchmark is a simple one involving a tank, a flat radiator slab, and a pipe connecting them. Heat flows between the tank and the slab. We may not know some of the parameters of this system, but we know the ending temperature of the slab. We can then use gradient descent optimization to best fit the system to the known data.</p><figure><img alt="A diagram of an equipment system consisting of a tank, a hot slab, and a pump circulating fluid between them." src="https://cdn-images-1.medium.com/max/1024/1*LhMr3KGjlTjD0W-UcQTR5Q.jpeg" /><figcaption>A schematic of the equipment system we are modeling for this benchmark.</figcaption></figure><p>In the case of this benchmark, we’ve removed the full gradient descent process. We focus only on measuring the time required to perform the original system model calculations (the forward pass) and the chained derivatives of all of the model calculations (the backward pass). We measure the total time it takes for both forward and backward passes over a set number of iterations, and record the average.</p><p>Because we needed gradient descent to perform our optimizations, we wanted to test various common technologies that had this capability. The most popular are the neural network training frameworks, like PyTorch and TensorFlow, which build in automatic differentiation as a library capability. These frameworks typically use Python as an interface language and C++ under the hood for performance.</p><p>We implemented the same calculation in differentiable Swift, PyTorch, and TensorFlow, and verified the numerical results for forward and backward passes to ensure they matched. I performed these benchmarks on an M1 Pro MacBook Pro, with all frameworks running purely on the host CPU. The following are the average timings for the combined forward + backward passes through the model, along with the relative slowdown of the other frameworks compared to differentiable Swift:</p><pre>+------------+------------+-----------+<br>| Version    |  Time (ms) |  Slowdown |<br>+------------+------------+-----------+<br>| Swift      |    0.03    |     1X    |<br>| PyTorch    |    8.16    |   238X    |<br>| TensorFlow |   11.0     |   322X    |<br>+------------+------------+-----------+</pre><p>Why is differentiable Swift so much faster in this benchmark? PyTorch and TensorFlow both excel at massively parallel calculations, especially when you can dispatch them on GPUs or other accelerators. However, in this case, we have only scalar calculations for the components of this heat transfer model, with the structure defined using types that express the component being simulated. Frameworks like PyTorch and TensorFlow rely on batching up parallel computation and dispatching that to accelerators or the CPU, and there’s overhead in that dispatch mechanism. In the Swift case, these calculations are compiled and run directly, without any dispatch overhead.</p><p>Like many current neural network frameworks, PyTorch and TensorFlow use Python as a host language. Python has performance disadvantages when compared to an ahead-of-time compiled systems language like Swift. The more time that a model spends in the high-level Python logic versus in the optimized C++ code under the hood, the greater this performance disadvantage becomes.</p><p>It’s for these reasons that this problem, which is very important for us to simulate, is well outside of the sweet spot for traditional neural network frameworks.</p><p>This is one representative case for where differentiable Swift can be significantly faster than alternatives, but is only one of many inside of our simulation and control systems. We believe there is a world of applications out there that have not won <a href="https://arxiv.org/abs/2009.06489">the software lottery</a> and are not well-served by current frameworks. We hope that differentiable Swift can enable those new applications.</p><p><a href="https://github.com/PassiveLogic/differentiable-swift-examples/tree/main/Benchmarks/BuildingSimulation">I invite you to try out these benchmarks for yourself</a>. Our examples repository has the full code for them, along with instructions for how to set up and run Swift, PyTorch, and TensorFlow environments for each benchmark variant. In addition, <a href="https://github.com/PassiveLogic/differentiable-swift-examples">the main repository</a> contains examples of how to use differentiable Swift along with documentation and detailed tutorials.</p><p><strong>Additional Swift benefits</strong></p><p>A fascinating aspect of Swift is that this speed doesn’t come at the expense of readability. Python is a language known for being easy to read, but it’s not common to say the same about a systems programming language. Compare the same function in our benchmarks written in Python (for PyTorch):</p><pre>def updateQuanta(quanta):<br>    workingVolume = (quanta[QuantaIndices.iflow] * dTime)<br>    workingMass = (workingVolume * quanta[QuantaIndices.idensity])<br>    workingEnergy = quanta[QuantaIndices.ipower] * dTime<br>    TempRise = workingEnergy / quanta[QuantaIndices.iCp] / workingMass<br><br>    resultQuanta = quanta + TempRise * torch.tensor([0.0, 1, 0, 0, 0])<br>    resultQuanta = resultQuanta * torch.tensor([0.0, 1, 1, 1, 1])<br><br>    return resultQuanta</pre><p>with the same in Swift:</p><pre>@differentiable(reverse)<br>func updateQuanta(quanta: QuantaType) -&gt; QuantaType {<br>    let workingVolume = (quanta.flow * dTime)<br>    let workingMass = (workingVolume * quanta.density)<br>    let workingEnergy = quanta.power * dTime<br>    let TempRise = workingEnergy / quanta.Cp / workingMass<br>    <br>    var updatedQuanta = quanta<br>    updatedQuanta.temp = quanta.temp + TempRise<br>    updatedQuanta.power = 0<br><br>    return updatedQuanta<br>}</pre><p>Developers can express the same concepts in a very readable manner in Swift code while maintaining strong type checking and value semantics. As an additional benefit, because we can differentiate the Swift code directly, there’s no need to cram variables into a tensor-like type to take advantage of automatic differentiation (like for the popular Python frameworks we’ve used here).</p><p>There are, of course, other capabilities that come with incorporating differentiability into a fast systems language. We can write our entire control platform, our simulations, and our control path-finding in the same language. It’s hard to say where one area of functionality begins and the other ends, unlike traditional machine learning deployments with hard boundaries between systems code and ML code.</p><p><strong>Ongoing optimizations</strong></p><p>It is a goal of ours to make differentiable Swift even faster. For example, two years ago <a href="https://medium.com/passivelogic/using-enzyme-autodiff-with-swift-afa1bc2dc102">we wrote an article</a> where we compared the speed of differentiable Swift against derivatives generated using the Enzyme framework. At that time, the Enzyme-generated derivative code was 477X faster than equivalent differentiable Swift. About a year later, <a href="https://github.com/apple/swift/pull/60520">we landed a significant optimization</a> for simple differentiable Swift functions that eliminated this performance gap, and now the Enzyme and Swift examples in that article are roughly equivalent in speed.</p><p>We’ve made significant improvements in the speed of differentiable Swift over just the last year. The same Swift building simulator we use here shows a <strong>2.4X improvement in performance</strong> when built using nightly Swift toolchain snapshots starting in January and running through September. We are currently prototyping an optimization that, in initial tests, leads to a further <strong>5.4X improvement</strong> over today’s nightly Swift toolchains. This totals a roughly <strong>13X speedup</strong> since the start of 2023.</p><figure><img alt="A chart of the increases in differentiable Swift performance on a simulator benchmark, starting at 1.0 in January and extending to 2.4X in September." src="https://cdn-images-1.medium.com/max/1024/1*6nCIqL2K9MJ2mFn4XvhvrA.jpeg" /><figcaption>The improvement in performance of our differentiable Swift simulator benchmark in nightly Swift toolchain snapshots over 2023.</figcaption></figure><p>In <a href="https://www.swift.org/blog/focus-areas-2023/#differentiable-swift">the published roadmap for differentiable Swift in 2023</a>, we described how less and less of our development effort would be needed to fix correctness issues and more time would then be devoted to optimizing the code generated by this language feature. We’re only starting to see the results of this, and I’m excited for what the next several months will bring.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=270da84f4fe4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/benchmarking-differentiable-swift-270da84f4fe4">Benchmarking Differentiable Swift</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[2023 LLVM Developers’ Meeting]]></title>
            <link>https://medium.com/passivelogic/2023-llvm-developers-meeting-7a04b29bc81c?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/7a04b29bc81c</guid>
            <category><![CDATA[pltech]]></category>
            <category><![CDATA[automatic-differentiation]]></category>
            <category><![CDATA[swift-programming]]></category>
            <category><![CDATA[compilers]]></category>
            <category><![CDATA[llvm]]></category>
            <dc:creator><![CDATA[Brad Larson]]></dc:creator>
            <pubDate>Fri, 08 Dec 2023 17:52:45 GMT</pubDate>
            <atom:updated>2023-12-08T17:52:44.893Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*s82NgwgzmJKsuCTwtwhtIw.png" /><figcaption>Members of the PassiveLogic compiler team at the 2023 LLVM Developers’ Meeting. From left: <a href="https://www.linkedin.com/in/kshit/">Kshitij Jain</a>, <a href="https://www.linkedin.com/in/viranchee/">Viranchee Ketan Lotia</a>, <a href="https://www.linkedin.com/in/manasij7479/">Manasij Mukherjee</a>, and <a href="https://www.linkedin.com/in/brad-larson-3549a5291/">Brad Larson</a>.</figcaption></figure><p>Recently, members of the PassiveLogic compiler team had the opportunity to attend the <a href="https://llvm.swoogo.com/2023devmtg/3815984">2023 LLVM Developers’ Meeting</a> in Santa Clara, CA. It was a tremendously rewarding experience for all of us.</p><p>For those who aren’t familiar, <a href="https://llvm.org/">LLVM</a> is a huge project that hosts compiler technologies which underpin a lot of the software and hardware products we use every day. Many modern programming languages are built upon the foundations that LLVM provides, such as Swift, Rust, and more. The LLVM Developers’ Meeting brings together experts from nearly every major technology company you’ve heard of to learn about advancements in the compiler, language, and related fields.</p><p>For the newer members of PassiveLogic’s compiler team, it was their first large industry and research conference and a valuable learning experience. They saw how the names behind big, important projects belonged to friendly people who loved to talk at length about their work. It’s always intimidating when you first arrive at a conference like this, but once you strike up that first conversation, you see how approachable everyone is.</p><p><strong>Differentiable Swift Roundtable</strong></p><p>PassiveLogic’s connection to LLVM is closely tied to <a href="https://medium.com/passivelogic/recent-advances-in-differentiable-swift-f60ccc0c6f7e">our work in advancing the Swift programming language</a>, particularly the experimental language feature of differentiable Swift. We’re working to make Swift the first systems language with first-class differentiation support integrated right into the language. This is in support of our <a href="https://quantumalliance.org/">unique physics-based digital twins</a> and their heterogeneous compute graphs, and what we see as a path to a post-deep-learning era beyond the confines of traditional homogeneous neural networks. To learn more about differentiable Swift, <a href="https://passivelogic.com/?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=launchevent#:~:text=Differentiable%20Swift">watch our PassiveLogic 2023 Launch Event.</a></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fxr8sV3GdhVk%3Fstart%3D3788%26feature%3Doembed%26start%3D3788&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dxr8sV3GdhVk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fxr8sV3GdhVk%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/dca17b50e210fbf449a77a814e4c8241/href">https://medium.com/media/dca17b50e210fbf449a77a814e4c8241/href</a></iframe><p>In addition to our immediate team, our partners from <a href="https://www.softek-toolchains.com/">Access Softek Toolchains</a> were present at the LLVM Conference as sponsors of the event. They do amazing LLVM work and have been critical to the advancement of differentiable Swift. Together, PassiveLogic and Access Softek Toolchains organized a roundtable on differentiable programming, and we had a great planning session about the future of <a href="https://medium.com/passivelogic/intro-to-differentiable-swift-part-2-differentiable-swift-25a99b97087f">differentiable Swift</a> as well as broader discussions about automatic differentiation in general. It was extremely productive for us to also connect directly with the Swift team members at the conference as we coordinated on next steps for this language feature.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QM2WdIRRn04CKeWN_yegxA.jpeg" /><figcaption>Our event T-shirts were a fun mix of gradient descent optimization and the slopes near our home office.</figcaption></figure><p><strong>MLIR</strong></p><p>Most of our team, myself included, opted to attend the pre-conference MLIR workshop. <a href="https://mlir.llvm.org/">MLIR</a> is a flexible framework for building compilers, growing out of the recognition that we are facing a Cambrian explosion of novel computing hardware and we need better ways of building software for this new world. A big theme of this year’s meeting was how MLIR is being adopted across a range of applications: from adversarial JavaScript detection, to <a href="https://circt.llvm.org/">circuit design via CIRCT</a>, to <a href="https://arxiv.org/abs/2202.04305">sparse tensor computation</a>.</p><p>I was particularly interested to hear about early explorations into how the <a href="https://enzyme.mit.edu/">Enzyme automatic differentiation framework</a> can be applied to MLIR. Enzyme is a fascinating technology for performing automatic differentiation — to date at the LLVM IR level — and <a href="https://medium.com/passivelogic/using-enzyme-autodiff-with-swift-afa1bc2dc102">we’ve previously written about our experiments with coupling this to the Swift language</a>.</p><p><a href="https://www.modular.com/">Modular</a>’s <a href="https://www.youtube.com/watch?v=SEwTjZvy8vw">keynote address</a> continued the theme of MLIR and its applications, presenting greater background on how they constructed their MLIR-based <a href="https://www.modular.com/engine">AI Engine</a> and in the process determined they needed a new language as an interface to that. <a href="https://www.modular.com/mojo">Mojo</a> was the result. I do have to say that I appreciated Jeff Niu from Modular doing a technical deep dive in front of everyone wearing a pink unicorn onesie. 🦄</p><p>You can catch an insightful conversation between PassiveLogic CEO Troy Harvey and Modular CEO Chris Lattner within our <a href="https://www.youtube.com/watch?v=xr8sV3GdhVk&amp;t=3876s&amp;utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=launchevent">2023 Launch Event here. </a>Both Modular and PassiveLogic share a vision of a future without boundaries between application logic and machine learning code. At PassiveLogic, this integration is necessary to solve real-world problems with our generative design technologies built on physics-based digital twins.</p><p><strong>Open Source Communities</strong></p><p>Beyond the technical discussions, I had a particular interest in the session held on the evolution of governance for the LLVM project. A full description of the proposal can be found in <a href="https://github.com/llvm/llvm-www/pull/54">this pull request</a>. LLVM brings together so many companies and groups across industry and research, all with their own (sometimes conflicting) interests. As a project, it has grown such that current processes can’t handle its scale. The members have considered recent governance discussions taking place across other sizable open source projects like Apache, Python, and Rust.</p><figure><img alt="A photograph of a slide from the LLVM governance discussion, highlighted by Bender from Futurama declaring he’s going to go build his own compiler." src="https://cdn-images-1.medium.com/max/857/1*P-1gslfJ6k6d0s2MP6E3AQ.jpeg" /><figcaption>Open source project governance sounds boring, but the discussion was both entertaining and important.</figcaption></figure><p>Particular mention was made of Nadia Eghbal’s book <a href="https://press.stripe.com/working-in-public">“Working in Public: The Making and Maintenance of Open Source Software”</a>, which I cannot recommend highly enough for anyone involved in open source projects.</p><p>Additionally, April Wensel gave a very well-received keynote about <a href="https://compassionatecoding.com/">compassionate coding</a>. She provided an easy-to-understand shorthand: “compassion is understanding plus action.” This rings true, and the healthiest open source communities that I’ve been a part of had policies of treating everyone with respect.</p><p>I want to personally thank <a href="https://www.linkedin.com/in/tanyalattner/">Tanya Lattner</a> and all of the organizers from the LLVM Foundation for their hard work in putting together the 2023 installment of the LLVM Developers’ Meeting. This was a unique opportunity for the compiler team at PassiveLogic to meet a ton of great people who are also working in this space. We came home with many areas to research, and encouragement from our LLVM peers to continue to push the limits of what can be achieved with compilers.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7a04b29bc81c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/2023-llvm-developers-meeting-7a04b29bc81c">2023 LLVM Developers’ Meeting</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Introducing Generative Autonomy. Autonomy for the Real World.]]></title>
            <link>https://medium.com/passivelogic/introducing-generative-autonomy-autonomy-for-the-real-world-844cdc38716e?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/844cdc38716e</guid>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[pltech]]></category>
            <category><![CDATA[autonomous-building]]></category>
            <category><![CDATA[generative-autonomy]]></category>
            <dc:creator><![CDATA[Troy Harvey]]></dc:creator>
            <pubDate>Tue, 28 Nov 2023 23:14:21 GMT</pubDate>
            <atom:updated>2023-11-28T23:14:21.385Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IytI431ssKeSnvNlf426jg.jpeg" /></figure><p>Our world operates inside of controlled systems, which fall into two distinct but unequal categories: transportation and the built world. The built world encompasses not just buildings but everything that goes on inside them. While the bulk of controlled systems operate within the built world, only cars have received any attention (and funding) from the world of autonomous systems. Yet there are no platforms for autonomy for the built world. Why? Because until now, the built world has been too topologically complex to make autonomous. To make the built world autonomous, you need autonomous systems that are generalized and can work with any set of systems.</p><p>During the PassiveLogic 2023 <a href="https://passivelogic.com/?utm_source=medium&amp;utm_medium=blog&amp;utm_campaign=launchevent">Launch Event</a>, we introduced the future of autonomy. Autonomy is more than single-purpose chatbots and single-purpose operations that require millions of dollars and months to train. True autonomy is flexible, dynamic, and capable of adapting to understand and manage any controlled system in a matter of minutes. True autonomy — by definition — requires generative autonomy.</p><p>During the keynote, I dove into generative autonomy and PassiveLogic’s breakthroughs making this groundbreaking technology possible. If you haven’t had the chance to watch the keynote, head to <a href="http://passivelogic.com/">passivelogic.com</a> to watch it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JVHd9EBz8DG4LsLuqjGxrQ.gif" /></figure><h4><strong>State of Play. No Real Autonomy Right Now.</strong></h4><p>With this incredible breakthrough, where does generative autonomy get applied first? The answer might not be obvious — but its application is imperative: It’s buildings. Buildings are the most complex robots in the world; the largest buildings trend towards <em>one million </em>input and output sensors and controllables. Buildings also consume more than 40 percent of our total energy consumption, and 40 percent of <em>that</em> energy is savable through better controls. The complexity and energy-saving opportunities of buildings make them ideal as the first application for generative autonomy.</p><p>So, let’s take a step back and look at the current landscape of autonomous systems to understand why buildings require more sophisticated and agile technology than any autonomous system on the market today.</p><p>The first industry to jump on autonomous systems was the automotive industry. But cars are actually pretty simple robots, once you get past vision in a chaotic environment. Autonomous cars are controlled by only a few dozen sensors or inputs and manage just three outputs: turning, braking, and acceleration. Once the AI system for one model has been perfected, it can be replicated and placed in car after car. The autonomous system is replicable and can only work in systems designed for it to succeed — but it’s brittle.</p><p>And if we look at the next big wave of automation, the chatbots and image generators dominating headlines today are fun, but they are actually pretty simple and only capable of one single task that requires years of training to pull off. They’re flashy, but not that sophisticated — and certainly shouldn’t be confused with AGI.</p><p>The real breakthrough in autonomous systems would be a system that would be general purpose and can easily and quickly be built to reflect the unique requirements of any system and work anywhere. That’s what can enable every unique one-off building to become autonomous.</p><h4><strong>Generative Autonomy. Autonomy for Everyone.</strong></h4><p>This next-generation AI system is required to enable autonomous buildings. As I said earlier, buildings are the world’s most complex robots, and this complexity renders them what is called in computer science an “intractable problem” — while theoretically solvable, it will never be solved through manual programming and static sequences. Add in the fact that every building is unique, and accounting for all possible building dynamics is nearly impossible. This reality is the root cause of the diverse challenges plaguing buildings today — from painstaking installation and poor system control to limited occupant comfort — leading to the world’s largest source of wasted energy.</p><p>Solving building complexity won’t come from point solutions like smart sensors, and solving their efficiency limitations won’t come from new “efficient” equipment installations. Current control platforms are designed to <em>limit complexity</em> because of the intractable nature of building control. Stacking an AI-optimizer on top doesn’t fundamentally solve any of those problems — it’s just another piece of technology to manage.</p><p>No one has ever before decided to embrace the inherent complexity of buildings. Until now. PassiveLogic has built the first brain for buildings that understands a building’s complexity and can simultaneously manage all of the systems within.</p><p>That means generative autonomy.</p><h4><strong>Four Breakthroughs. 17 Products. One Solution.</strong></h4><p>We’ve built the technology for generative autonomy with four key breakthroughs and 17 products that put generative autonomy in your hands:</p><ol><li><strong>A Complete Ecosystem of Digital Twin Applications</strong></li></ol><p>🔍 Quantum Lens</p><p>🏢 Building Studio</p><p>🦾 Autonomy Studio</p><p>⚛️ Quantum Creator</p><p>🪪 Quantum Passport</p><p>🧭 Quantum Explorer</p><p><strong>2. A Breakthrough Platform for Generative Autonomy</strong></p><p>🐝 Hive &amp; Hive Mini</p><p>🍯 Cell Family</p><p>👆 SenseTouch and SenseNano</p><p><strong>3. The First Digital Twin Language for Composable AI</strong></p><p>👯‍♂️ Quantum Digital Twin</p><p>🤖 QuantumAPI</p><p><strong>4. The World’s Fastest AI-Language and Tool Chain</strong></p><p>🪽🔥 Differentiable Swift Compiler</p><p>Over the coming weeks, I’ll be sharing blogs that go deeper into these breakthroughs and our products, sharing additional insights and conversations with building controls, real estate, and deep-learning technology leaders as they discuss how PassiveLogic’s technology will transform their industries.</p><p>People sometimes say that we’re like 10 start-ups in one. They’re absolutely right. You can’t transform the built world with point solutions alone.</p><p>Follow me on <a href="https://www.linkedin.com/in/troy-aaron-harvey/">LinkedIn</a>. Join our <a href="https://passivelogic.com/">waitlist</a> for our software product release in December 2023.</p><p>Follow PassiveLogic on <a href="https://www.linkedin.com/in/troy-aaron-harvey/">LinkedIn</a> and <a href="https://www.youtube.com/@passivelogic_/videos">YouTube</a> to stay caught up on the latest news.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=844cdc38716e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/introducing-generative-autonomy-autonomy-for-the-real-world-844cdc38716e">Introducing Generative Autonomy. Autonomy for the Real World.</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What’s it like to be part of a generative AI startup?]]></title>
            <link>https://medium.com/passivelogic/whats-it-like-to-be-part-of-a-generative-ai-startup-fc44a2fb81ed?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/fc44a2fb81ed</guid>
            <category><![CDATA[generative-ai]]></category>
            <category><![CDATA[sustainability]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Tanner Wolfe]]></dc:creator>
            <pubDate>Mon, 28 Aug 2023 22:18:19 GMT</pubDate>
            <atom:updated>2023-08-28T22:41:03.139Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qHKTc7eHsi1gpiqkUZREkw.jpeg" /><figcaption>Middle: Mahan Tabatabaie. Clockwise from top: Saugat Tripathi, Eden Harvey, Rachel Molen, An Dang, Geoffrey Turbeville, Aidan Terlizzi.</figcaption></figure><p>When building an ecosystem of hardware and software products that enable generative autonomy, it’s all hands on deck. Every line of code and every unit test is a necessary piece to achieving the goals in <a href="https://passivelogic.com/">PassiveLogic</a>’s product architecture (the masterplan, if you will.)</p><p>Achieving this technically-intensive effort requires focused contributors who take on important projects within our architecture. PassiveLogic provides interns the opportunity to play a part in developing this technology. Below, we share insights from our latest cohort of interns and their stories about working at PassiveLogic.</p><h3><strong>Internships that Matter</strong></h3><p>PassiveLogic has developed great breadth across multiple engineering and design disciplines that enables us to build the first platform for generative autonomy. Interns work across that breadth, participating in our teams and adding to our momentum. This is made possible through identifying stretch projects for the intern that fulfill consequential work streams within our overall platform development.</p><p>For <a href="https://www.linkedin.com/in/aidanterlizzi/">Aidan Terlizzi</a>, finding an internship where he could get involved in impactful development work was one of the important factors while job searching. “During the COVID-19 pandemic, I taught myself how to code outside of school. Every summer I’ve been doing software engineering internships to gain experience,” said Aidan, who is in his senior year at Rice University, completing a B.S. in Chemical and Biomolecular Engineering. “It’s rare to find a company that is an amalgamation of science and software. This was perfect for me because I wanted to find an internship that would incorporate the valuable concepts I learned during my degree, as well as the software engineering skills I learned in my own time.”</p><p>Prior to his internship at PassiveLogic, Aidan created an AI tool that used natural language processing to carry out automatic edits on college essays as they were being written. At PassiveLogic, Aidan uses similar methods to build a tool that creates generative digital twins using large language models (LLM). The tool can generate a full 3D floor plan according to chat prompts such as “build a house that has on the first floor: a living room, dining room, and three bedrooms.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l8JvE28vBPulmbiCRM6SpA.jpeg" /><figcaption>Left to right: Aidan Terlizzi, Majid Karami, Bhavik Rathod, Kavin Sundarnath. J</figcaption></figure><p><a href="https://www.linkedin.com/in/saugat-tripathi/">Saugat Tripathi</a>, our BLE Wireless Intern, found PassiveLogic’s job description intriguing as it mentioned working with “new kinds of wireless comms and algorithms.” This aligned with his M.S. in Electrical and Computer Engineering at Miami University, where his thesis focused on wireless communication, machine learning, and deep reinforcement learning for optimizing wireless comms networks using drone-based communication systems. At PassiveLogic, he currently works on BLE Mesh based algorithm development.</p><p>For <a href="https://www.linkedin.com/in/geoffrey-turbeville-aa7759225/">Geoffrey Turbeville</a>, getting an internship where he could exercise his skills from an extensive career in construction and his engineering educational background was a rare find. Geoffrey recently completed his Bachelors of Engineering in Mechanical Engineering at The City College of New York. At PassiveLogic, Geoffrey blends his hands-on field experience with education as an Applied Building Scientist and Field Service Technician. His unique background made him a perfect fit to traverse both roles, which involves maintaining PassiveLogic’s alpha site testing grounds and analyzing their performance.</p><p>As part of the Verification and Validation team, Geoffrey’s work is essential to getting our alpha sites/external laboratories ready for the next phase of field testing.</p><h3><strong>Developing a Continuous Learning Mindset</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6Lj4tE-jNJOWXuOsg-KIvg.jpeg" /><figcaption>Left to right: Rachel Molen, An Dang, Eden Harvey, Shishir Yalburgi</figcaption></figure><p>We often tell our employees that “your success is only limited by your enthusiasm.” Our internships are designed to provide dedicated interns an opportunity to transition into full-time employment. We aim to set our interns up for success by instilling in them an understanding that real growth is not measured by regimented management tiers, but by continuously learning, trying new things, and perfecting your craft. In fact, being able to enthusiastically take part in uncharted territory is not just a “nice to have”, its something we’re looking for in every person who applies for a job at PassiveLogic.</p><p>Our interns hold this mindset of innovation and growth. <a href="https://www.linkedin.com/in/rachmolen/">Rachel Molen</a> is an intern on our AI team who recently graduated with a B.S. in Applied and Computational Mathematics Emphasis and a Minor in Music at Brigham Young University. Within her first 3 months here, Rachel created a tool that visualizes optimization simulations. In a building, there are a lot of high-dimensional equations and functions going on, so Rachel’s tool visualizes those dimensions in a meaningful and clear way. “Other PassiveLogic engineers can use the tool I created to see what the cost function looks like in lower dimensions. This can be especially helpful when dealing with gradient descent parameter-tuning, as gradient descent sometimes picks local minima over global if we start the algorithm at certain coordinates or set the step-size too large.”</p><p><a href="https://www.linkedin.com/in/mahan-tabatabaie-74896597/">Mahan Tabatabaie</a> is another intern who joined the AI team. As an Autonomous Systems Intern, Mahan’s multidisciplinary role brings AI, thermodynamics, and mechanical engineering together. While at first there was a lot to take in, Mahan credits the guided onboarding process that helped him get started. “If I had started to code from day one, I wouldn’t have a clear understanding of our project. I was given the Ecosystem Design Document so I could get up to speed. I read the whole thing, all 60 pages, and it was the perfect thing for me to get an idea of what we do, our goal, and how these products work together. It helped me to get an understanding of the big picture and what my role in that is.”</p><p>After onboarding, Mahan was able to start on his project. Mahan is currently in his 3rd year of his Ph.D in Computer Science at the University of Connecticut, focusing on driver mobility. “In my Ph.D, I analyze driver behavior, safety, and autonomous vehicles. At PassiveLogic, I was given the project to apply human mobility to the indoor positioning of people and assets within a building,” Mahan explains. This is a new dimension for Mahan, who utilizes GPS signals in his doctorate research. At PassiveLogic, he is using other signals such as WiFi or BLE, because GPS signals can’t pass through the walls of a building. “This is all new to me, but what’s interesting at PassiveLogic is that there’s a physics-based digital twin of the building which makes this problem so much easier to solve. We know exactly what the walls are made of, so we can predict how the signal passes through the wall. With this information, we can simulate exactly what would happen to the signal, and locate it more accurately.”</p><h3><strong>A Principled Approach to Engineering</strong></h3><p>Ultimately, the goal of our internship program is to instill a principled approach to engineering that sticks with all of our interns throughout all of their projects. This mindset is what sets us apart from other technology firms.</p><p>Having experience in a real work setting helped Eden Harvey to practice his own principled approach to engineering. “In school, you learn the engineering basics and apply it to a set problem that’s been designed to be solved the way you learned earlier that day. But in reality, when you’re faced with a problem, there’s not a clear path to solving it.”</p><p>Eden is a Quantum Lens Testing Intern on the Verification and Validation team, currently pursuing a B.S. in Computer Engineering at the University of Utah. “The most valuable thing I’ve learned here is reorientation of that problem-solving process. We approach problems with the widest aperture possible,” Eden explains. “If you start with a broad knowledge-gathering strategy, and narrow things down from there, you’ll be better set up to solve the problem. A big thing students run into is that they may have 10 different directions they could take a problem, and they go all the way down on one pathway.” This can result in lost time if its not the right solution. Instead, Eden spends time taking different approaches, and then narrows down the solutions to reach his goal.</p><p><a href="https://www.linkedin.com/in/an-dang-91b7411b8/">An Dang</a> has learned a similar approach. An is beginning his M.S. in Robotics at the University of Michigan, and is currently interning on our AI team. “Everything I did in my research was in a controlled setting. In a controlled setting, you can simplify a lot of things to make an idea happen. But in a real production setting, you can’t just assume that you have a magical building that adheres perfectly to what you expect, because it has to adhere to the laws of physics,” An explains. “Developing a real product for the market teaches you how to approach things that aren’t so clear cut.”</p><p>Mahan Tabatabaie agrees. “My mentor, Porter Child, helped me to see AI and software engineering with a different perspective. He taught me to develop my vision first. So I always start with considering all the possible conditions that could occur in the future. I think about the long-term design: what could happen to this product in a few months? A few years? 10 years?”</p><p>It’s important that everyone on our team, whether they’re an intern or full-time employee, is thinking of the big picture impact through every step of the development journey. We’re working hard to expand the paradigm for what is possible through engineering, so this long-term thinking has been fundamental as we release our fully autonomous platform to the market.</p><blockquote>We’re looking for people who are committed to lifelong learning and are driven by a love of what they do. Check out our latest job openings on <a href="https://passivelogic.com/careers/">PassiveLogic.com/Careers</a>.</blockquote><blockquote>We know there are candidates who might not fit everything outlined in a job description, or have experience and skills we haven’t considered. If that’s you, talk to us at careers@passivelogic.com.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fc44a2fb81ed" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/whats-it-like-to-be-part-of-a-generative-ai-startup-fc44a2fb81ed">What’s it like to be part of a generative AI startup?</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Guide to Verification and Validation]]></title>
            <link>https://medium.com/passivelogic/guide-to-verification-and-validation-b3698dd42cb3?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/b3698dd42cb3</guid>
            <category><![CDATA[qa]]></category>
            <category><![CDATA[verification]]></category>
            <category><![CDATA[validation]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[pltech]]></category>
            <dc:creator><![CDATA[Jeremy Fillingim]]></dc:creator>
            <pubDate>Wed, 16 Aug 2023 21:37:06 GMT</pubDate>
            <atom:updated>2023-08-16T22:20:22.370Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*GYf5RZayJ375WYmpW3r54g.jpeg" /><figcaption>John Easterling and Doh Haroldsen (left to right)</figcaption></figure><blockquote>“If you don’t have time to do it right, when will you have the time to do it over?” John Wooden, legendary UCLA basketball coach</blockquote><p>At <a href="https://passivelogic.com/">PassiveLogic</a>, a key component of our product evaluation process is asking ourselves, “Have we met the objectives of the original design?” If we develop tools that enable people to solve problems in an elegant and straightforward way, then we’ve done our job. If not, we go back to the drawing board.</p><p>The tool we’re building is a fully autonomous platform of hardware and software products that solve the all-too-common pain points that the building controls industry faces today. The platform aims to eliminate the negative aspects of siloed proprietary systems, outdated technology, and fixed programming, all the while optimizing for energy efficiency and occupant comfort through generative autonomy.</p><p>Generative autonomy enables edge-based, real-time decision making via AI-powered controls generated using physics-based digital twins. This is game changing for the market: we’re upping the ante by developing an entire end-to-end ecosystem that combines the best of BIM, generative design, and differentiable computing.</p><p>Building a platform that can deliver on its promise and be adopted by a typically risk-averse industry is tough. But like John Wooden famously said, “If you don’t have time to do it right, when will you have the time to do it over?” It requires a lot of planning, developing, testing, and then going back to square 1 if we don’t have things exactly right. A fundamental part to making this possible is our Verification and Validation team, which acts as an evaluator to the rest of our engineering teams.</p><h3><strong>A fully-fledged testing operation</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S7ZKVHhoXC_Bw0LnG59EJw.jpeg" /><figcaption>Our team members in India: Islam Ahmad, Paritosh Kumar, Raj Kumar Singh, Sulekha Malik (left to right)</figcaption></figure><p>We call our team the Verification and Validation team because “testing” on its own doesn’t quite cover the complexity of this process. <strong>Verification</strong> should answer the question “Are we building the product correctly?” On the other hand, <strong>validation</strong> answers “Are we building the right product?” Our goal is not only to meet, but also reset customer expectations. Our stakeholders aren’t familiar with autonomous control, so we must prove to the market that our AI is safe, robust, and easy to use, so that people make the leap to adopt it.</p><p>To take full advantage of the benefits of a complete testing process, we’ve invested in a dedicated Verification and Validation team so that we can bring specialized expertise straight into our offices. With a team that is not involved directly in the development process, they’re better able to approach products objectively, which helps them to identify issues that may be overlooked by those who are familiar with the implementation.</p><p>To cover the entire testing process, we’ve divided the V&amp;V team into several test groups:</p><blockquote><strong><em>Software Testing and Q/A:</em></strong><em> This group tests all embedded systems through unit testing, integration testing, and system validation testing.</em></blockquote><blockquote><strong><em>Hardware Test: </em></strong><em>This group tests the physical and mechanical capabilities of our products, as well as the accuracy and precision of all sensors.</em></blockquote><blockquote><strong><em>User Software Test: </em></strong><em>This group designs test cases as per the specific requirements of our user software applications, and then validates it through all phases of testing. They perform daily sanity testing on every nightly build, and report those results to the stakeholders. They act as a gatekeeper, providing the final sign-off on testing before every release.</em></blockquote><blockquote><strong><em>Applied Systems: </em></strong><em>This group designs field tests to see how our products fare in the scenarios they’re designed for. They also ensure that all pieces of the puzzle come together through “Integration testing”. One product might work well by itself, but it is vital that the entire platform works well.</em></blockquote><p>As a whole, the Verification and Validation team possesses expertise across testing methodologies, techniques, and tools. Our team members are skilled in designing comprehensive test plans, creating test cases, and executing various testing activities. Their collective experience across software and hardware enables them to uncover potential issues and ensure the products meet the required quality standards. “We’ve been lucky to have hired a whole lineup of incredible test engineers, each of whom brings their own set of highly sought-after skills,” <a href="https://www.linkedin.com/in/john-easterling-78606031/">John Easterling</a>, Director of Verification and Validation, remarked. “Having this dream team means that we have people who allocate their time and resources to thoroughly testing the software, and thus provide the utmost safety and security of data for all of our users. This leads to higher customer satisfaction and a seamless user experience.”</p><h3><strong>Designing the ultimate testing process</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QanvTgEU15KrzCGTAMuSMA.png" /><figcaption>Xingsheng Wei and Christian Graff (left to right)</figcaption></figure><p>We undertake a Design-Build-Execute strategy when we carry out our testing. The first step is communicating with the design and product management team, who are primarily responsible for creating the detailed technical specifications needed for each product release and aligning them to market needs. After we’ve met and discussed the product requirements, we <strong>design</strong> a test plan that consists of individual test and test fixture definitions. We then <strong>build</strong> the rigorous tests and “fixtures” or “jigs” that can successfully carry out the tests. Then we <strong>execute</strong>. These test suites verify product architecture, implementation, and functionality against those initial requirements.</p><p>The name of the game is to break the product every way we can, so that we can ensure that the user will never have to experience those breakages themselves. Like <a href="https://www.linkedin.com/in/cory-mosiman-9b1727101/">Cory Mosiman</a>, Applied Systems Validation and Software Engineer, often says “They make it, we break it.” While these tests are customized by each speciality test group, we always ask ourselves a long list of key questions throughout the process that might include:</p><blockquote><em>Does the product perform in an any </em><strong><em>unexpected</em></strong><em> ways?</em></blockquote><blockquote><em>Does the product achieve our </em><strong><em>objectives</em></strong><em>?</em></blockquote><blockquote><em>How does the product perform in a </em><strong><em>virtual</em></strong><em> </em><strong><em>environment</em></strong><em>, and does that track to how it behaves in the </em><strong><em>real</em></strong><em> </em><strong><em>world</em></strong><em>?</em></blockquote><blockquote><em>If it does achieve the objectives, to what degree, and </em><strong><em>in what ways does it succeed</em></strong><em>?</em></blockquote><blockquote><em>If the product fails the test, to what degree, and </em><strong><em>why</em></strong><em>?</em></blockquote><h3><strong>Why do we perform validation and verification tests?</strong></h3><p>In order to disrupt an industry, carefully designing and carrying out a battery of tests is necessary to provide a truly effective product. Our User Software Test Lead, <a href="https://www.linkedin.com/in/paritosh-k-9246021a/">Paritosh Kumar</a>, who previously led testing efforts at Johnson Controls, Aptara, and IBM, is a huge proponent of pushing the product to be as effective as possible. “All of these processes, though lengthy, contribute to the development of robust, high-quality systems, products, and models,” he explains. He expands on the main benefits to a fully-fledged testing process below:</p><blockquote><strong><em>Error Detection: </em></strong><em>By subjecting the system to different scenarios and inputs, we’ll uncover potential issues, bugs, and flaws that may arise during real-world usage. This allows for debugging and refining the system before it gets deployed.</em></blockquote><blockquote><strong><em>Performance Evaluation: </em></strong><em>We can measure accuracy, precision, recall, speed, and efficiency, to determine how well the system or model performs under different conditions. This evaluation helps make informed decisions about its readiness for deployment or further improvements.</em></blockquote><blockquote><strong><em>Risk Mitigation: </em></strong><em>Perhaps the most crucial part of the process is that by identifying and fixing issues, we reduce the chances of encountering problems in the field. This helps minimize potential financial losses, reputation damage, or safety hazards that may arise from using an untested system.</em></blockquote><blockquote><strong><em>Iterative Improvement: </em></strong><em>By collecting feedback and insights from the testing phases, we can make necessary adjustments, optimize performance, and enhance the overall quality of the system or model.</em></blockquote><h3><strong>Failure is not an option</strong></h3><p>In the end, our goal is to make sure that our product works for each individual user and they have an extraordinary experience in every interaction they have with our products. Cory Mosiman, who designs and builds field tests for our products, previously worked at the <a href="https://www.nrel.gov/">National Renewable Energy Laboratory</a> where he was key contributor to the <a href="https://github.com/NREL/alfalfa">Alfalfa</a> project. His valuable experiences in applied research have given him clear understanding that when it comes to bringing AI to the industrial world, there are a lot of risks, costs, and operational maintenance involved. He likes to think of NASA Flight Director Gene Kranz, who famously said “failure is not an option” in regards to the Apollo 13 moon landing. “During the space race, getting into space was a new concept where they needed to define testing procedures to ensure that the launch would be successful,” Cory explained. “We might not be going into space, but we’re applying that same methodology to making sure that our product is completely safe, completely vetted.”</p><p>Tests are designed using the appropriate apparatus to simulate use cases as similar as possible to the real world. For example, to validate a device like PassiveLogic’s Sense Nano, we vet the hardware through environmental limit tests in a controlled temperature / humidity chamber and measure how our sensors perform against reference sensors across the operating range to check if our product performs at high accuracy. “Every time we design our tests, our goal is to check — is the product safe? Is the product stable? Does it meet our documented specs? And does it meet the client’s needs?” <a href="https://www.linkedin.com/in/graffchristian/">Christian Graff</a>, Hardware Test Engineer, said.</p><p>Everything is driven by the design. “If the product doesn’t meet the objectives we’re not afraid of sounding the alarm and bringing the news to the designers. We work closely with the development teams to provide feedback on how they will need to refine the product to make sure it meets PassiveLogic standards,” says John Easterling. This feedback can seem critical, but meeting our original requirements is crucial in guaranteeing an excellent, consistent product.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b3698dd42cb3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/guide-to-verification-and-validation-b3698dd42cb3">Guide to Verification and Validation</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Onboarding, Elevated: The Product Demo Day]]></title>
            <link>https://medium.com/passivelogic/onboarding-elevated-the-product-demo-day-ec2bea50684d?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/ec2bea50684d</guid>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[sustainability]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[organizational-culture]]></category>
            <dc:creator><![CDATA[Tanner Wolfe]]></dc:creator>
            <pubDate>Fri, 28 Jul 2023 14:32:19 GMT</pubDate>
            <atom:updated>2023-07-28T14:37:14.726Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rw4KTLnGecNDvs8HF1rIXQ.png" /></figure><p>PassiveLogic has grown from a small group of multidisciplinary engineers and high-performance building scientists to an international team backed by top technology and real estate investors including <a href="https://techcrunch.com/2022/09/20/autonomous-building-control-startup-passivelogic-inks-partnership-with-nvidia-secures-15m/">NVIDIA</a> and <a href="https://venturebeat.com/ai/passivelogic-raises-15m-for-autonomous-building-controls/">Brookfield</a>. As we continue to grow beyond our 100 employee milestone, the demands of collaboration are greater, and it becomes even more important that we develop processes to stay aligned.</p><h3><strong>Connecting to the bigger picture</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Hhqjw-HT0TL2Iws4y6piKQ.jpeg" /></figure><p>PassiveLogic’s accelerated growth is directly correlated with each employee’s commitment to the company vision: to empower people through generative autonomy to solve the world’s largest climate challenges. <a href="https://passivelogic.com/careers/">We attract top talent</a> from all over the world — determined, innovative, and energetic engineers, researchers, and creatives who care deeply about making disruptive technology.</p><p>What’s been clear to us is that when our employees understand how the outcomes of their work connect to the bigger ecosystem of technology and products that we’re building, they produce better work.</p><p>One of the most successful practices we’ve prioritized to help our employees cultivate this awareness is our “Product Demo Day” which provides all new members of our team (and anyone else who would like to get a refresher) a fun, interactive series of guided demos on each of our products in our autonomous platform.</p><h3>Collaboration is key</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gN1i7PHN_U1PRPO-aTpTMw.jpeg" /></figure><p>Building a fully autonomous platform is hard work. It involves a lot of moving parts to develop an entire ecosystem of hardware and software products. In order to succeed here, employees will need to get up-close and personal with the technology developments their teammates have contributed to and gain a deep perspective of how their own work fits in the bigger picture.</p><p>Led by the team leads who worked on the products themselves, the Product Demo Day illustrates how each of the products work and how they solve commonly-experienced user pain points. Because each employee is an important contributor to our vision, everyone, no matter their technical background, gets an opportunity to understand what our users care about. This helps employees to easily chart a logical path from the project they’re working on towards the user experience.</p><p>The Product Demo Day highlights our cross-functional collaboration across teams — a valuable hallmark of our culture. When Mobile Team Lead <a href="https://www.linkedin.com/in/adam-petersen-b7a30a116/">Adam Petersen</a> started at PassiveLogic, he realized that the open-door policy helped him to grow quickly: “We’re not siloed in ‘this is what you do and that’s all you’re ever going to know about.’”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YMupavAGcOxkP-ny9A2R-g.jpeg" /></figure><p>Here, we encourage our engineers to step outside their usual office spaces to help each other, ask questions, and share their skills. Because of this, we’ve nurtured an environment that generates user-centric products.</p><p>We like to emulate the ethos that “the whole is greater than the sum of its parts.” Even when we were just a small group of employees, one of the things that made our team so strong was the collective understanding that transparency and collaboration is fundamental to our success and growth.</p><h3><strong>Building an entire ecosystem in tandem</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E5m-I4c3FTgAgE8oZuDGMQ.jpeg" /></figure><p>Developing an ecosystem of products is no simple feat. One of our software team leads, <a href="https://www.linkedin.com/in/alec-nelson-02873287/">Alec Nelson</a>, said: “The Demo Day is hugely important for PassiveLogic employees — it’s the fastest and most beneficial way to understand as a whole what the company is doing.” Alec knows the learning curve is steep for new employees. “Getting oriented can get complicated, so having a day where you get an in-depth explanation of the products is the perfect onboarding experience for someone to see how all of these products interact.”</p><p><a href="https://www.linkedin.com/in/caroline-genster-76a144114/">Caroline Genster</a>, PR and Media Relations Strategist, started at PassiveLogic earlier this year. “At the beginning, it was information overload. The technology we’ve built here is so advanced.” She continued, “Getting to actually see it and experience it is a super helpful way for me to conceptualize how all the pieces come together.”</p><h3><strong>A superior user experience</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RjSssvjkUHYOEh0lzNg7jA.jpeg" /></figure><p>The Demo Day is also a valuable opportunity to showcase how we’ve made progress on our initial design plans and gain feedback from our peers. We’ve carefully designed our products to meet user needs, so this is a chance to put that to the test and onboard our employees, too. “Everybody has a love for the projects they’re working on,” said <a href="https://www.linkedin.com/in/joel-j-52261a101/">Joel Joseph</a>, AI Application Engineer. “It’s pretty rare to see that. Most people at other companies collect their paycheck and go. But here, my favorite part is each team’s pure excitement to share the tangible outcomes of their work with newer employees.”</p><p>Sharing our developments in real time in front of real users means that it opens the product to critique. We’ve essentially established a live feedback loop that helps us to add new features or fix bugs we didn’t see. Adam has received many helpful suggestions to help push the product: “Some of the greatest ideas come from sharing your work and receiving feedback.”</p><p>Alec has also improved in his understanding of the product itself. “Explaining what our products do helps me to build my understanding of the tool to a deeper degree,” he explained. “As we’ve done more of these demos, I’ve realized that it doesn’t feel like a presentation anymore. It’s evolved into a freeform discussion where I can learn more about my user and then paint a picture that is customized to them.”</p><p>Keeping the end-users’ needs in mind through every step of the way is a fundamental value at PassiveLogic — it’s something we want to instill in our newer employees from Day 1. Thanks to the collaborative efforts of our teams and the live feedback they receive during the product demos, we build better products.</p><p>_</p><p><em>If you’re interested in working at PassiveLogic, check out </em><a href="https://passivelogic.com/careers/"><em>our careers page.</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ec2bea50684d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/onboarding-elevated-the-product-demo-day-ec2bea50684d">Onboarding, Elevated: The Product Demo Day</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[15 Years of the iPhone App Store: The Story of Molecules]]></title>
            <link>https://medium.com/passivelogic/15-years-of-the-iphone-app-store-the-story-of-molecules-f0acd07e44a7?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/f0acd07e44a7</guid>
            <category><![CDATA[pltech]]></category>
            <category><![CDATA[app-store]]></category>
            <category><![CDATA[apple]]></category>
            <category><![CDATA[iphone]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Brad Larson]]></dc:creator>
            <pubDate>Thu, 20 Jul 2023 15:01:36 GMT</pubDate>
            <atom:updated>2023-07-20T15:01:36.674Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l3PsA4j6x7GEIb3SasGO4g.png" /></figure><p>The iPhone App Store launched 15 years ago, and transformed the way that we make, use, and obtain software. More personally, it had a tremendous positive impact on my life. I had one of the first 500 applications that launched with the store that day, an open source 3D molecular modeler called <a href="http://www.sunsetlakesoftware.com/molecules/index.html">Molecules</a>.</p><p>The reception of the Molecules app went far beyond my wildest dreams. I made many friends, met my amazing wife, and took my career in a completely different direction, all as a result of this one little project.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*byJm_Us5CMhfaleQKbf3VA.png" /></figure><p><a href="http://www.sunsetlakesoftware.com/molecules/index.html">Molecules</a> was an application for iPhone (and later iPad and Mac) that let you render and manipulate full 3D models of molecular structures right on your mobile device. While we take advanced 3D graphics on our phones for granted nowadays, at the time this had never been done before. Molecules also let you search for and download biomolecules from the <a href="https://www.rcsb.org/">RCSB Protein Data Bank</a>, as well as small molecules from the NIH’s <a href="https://pubchem.ncbi.nlm.nih.gov/">PubChem</a>.</p><figure><img alt="An image of the iPad interface to Molecules, with a list of molecular structure names on the left and a 3-D visualization of a transfer RNA molecule on the right." src="https://cdn-images-1.medium.com/max/1024/1*Y0UOen_L_Y2xA9RM3lABEA.png" /><figcaption>The interface for Molecules on iPad, showing a 3D rendering of a transfer RNA molecule.</figcaption></figure><p>It all came about due to a request by my brother. At the time he was a researcher working toward his Ph.D. in protein crystallography, and had just come back from a poster session. I heard him complaining about how you couldn’t really show off a new 3D protein structure on a flat printed poster, and he couldn’t hold a laptop in his hands through the entire poster session. There had to be a better way to do this.</p><p>I’ve been excited about handheld computing since the <a href="https://en.wikipedia.org/wiki/Pilot_1000">U.S. Robotics Pilot</a>, and as soon as I bought the first iPhone, I jailbroke it to begin developing with the private Cocoa Touch interface on the device. I was floored by <a href="https://www.youtube.com/watch?v=IS5bMRKc5Ss">the announcement of the beta iPhone SDK</a>, especially the previously unknown OpenGL ES capabilities of the device. At Apple’s Worldwide Developers Conference in 2008, while watching a presentation on how OpenGL ES worked on the iPhone, I started thinking more about my brother’s problem and wondered if I could actually pull off a molecular renderer on this little handheld computer.</p><p>I came back from the conference and immediately dove in, devoting every night and weekend to learning OpenGL ES and building prototypes. I started believing this could actually work, and rushed to get an application ready before the App Store submission deadline. It was both exhilarating and nightmarishly stressful to be one of those first developers working with the beta iPhone SDK before launch. For how unbelievably broken some things were back then, all of us early developers got our apps built, signed, and submitted through sheer force of will.</p><p>It’s hard to describe the pride I felt at seeing my little app’s name and icon on the App Store when it went live. I had made something that anyone in the world could download and use. And they did, in numbers I couldn’t have imagined. Molecules was featured on the front page of the App Store multiple times, downloaded over 3.5 million times before I stopped getting statistics (we have much better information in the modern App Store), and even hit #1 in the Education category for a little while:</p><figure><img alt="The Top Free Apps chart from iTunes, with Molecules at #1 above the NASA App." src="https://cdn-images-1.medium.com/max/226/1*WO8lrhEBvQtCojY29PdJDg.png" /><figcaption>Molecules made it (briefly) to the top of the Free Apps in the Education category on iTunes.</figcaption></figure><p>I still smile when I see the icon for it in the original Apple ads for the iPhone 3G:</p><figure><img alt="A frame from one of the original iPhone 3G ads, showing the icon for Molecules on the home screen." src="https://cdn-images-1.medium.com/max/839/1*CdHWE2JVFxdo-07mjeNoJw.jpeg" /></figure><p>After a little while, Apple announced another new handheld device: the iPad. That larger canvas seemed like the perfect place for a molecular modeler, so I once again worked hard to adapt my iPhone application to this new platform. It was there at the launch of the iPad App Store, and even became one of the standard apps loaded on demo iPads at the Apple retail stores. I can’t tell you how honored I was to see it on the late Steve Jobs’ iPad on stage:</p><figure><img alt="A capture from Steve Jobs running an onstage demo with Molecules on his iPad." src="https://cdn-images-1.medium.com/max/889/1*Ls05sOnxzASM5MUqCZ4lkA.jpeg" /></figure><p>At one point, I was approached by the experienced educators behind the company <a href="https://digitalworldbiology.com/">Digital World Biology</a> who wanted to try for a grant to expand Molecules into a real teaching tool for classrooms across the world. The grant was funded by the <a href="https://www.nsf.gov/">NSF</a>, and they brought on a brilliant local iPhone developer named Janie to work with me on the project. She and I got along terrifically, and after the grant was done, she joined my company to work on robotic control software. We eventually realized that we brought out the best in each other, and Janie and I will have been married for five years this September.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rc_mQ9wo12YG-u2mWJUm-g.png" /></figure><p>The iPad also had new hardware rendering capabilities, going beyond the fixed-function pipeline of OpenGL ES 1.0 to the programmable shaders of OpenGL ES 2.0. This opened up whole new rendering techniques, and I became determined to adapt <a href="https://pubmed.ncbi.nlm.nih.gov/17080857/">a recently-published technique by Tarini and coworkers</a> to mobile devices. That was <a href="http://www.sunsetlakesoftware.com/2011/05/08/enhancing-molecules-using-opengl-es-20/index.html">an incredibly fun and educational journey</a>, ending up in <a href="http://www.sunsetlakesoftware.com/sites/default/files/NanotechMoleculesPaper-Submitted_0/index.pdf">a conference paper</a> and a nice mention onstage in WWDC 2011’s session “What’s New in OpenGL ES”.</p><figure><img alt="An image showing an old OpenGL ES 1.0 rendered molecule on the left and a much more realistic OpenGL ES 2.0 rendered molecule on the right." src="https://cdn-images-1.medium.com/max/1024/1*aDagIfgjsoJDEyjKR9thVg.png" /><figcaption>The old OpenGL ES 1.0 renderer (left) and the extremely realistic new OpenGL ES 2.0 renderer (right).</figcaption></figure><p>I soon went deep down the rabbit hole of GPU-accelerated image processing, which first took me to machine vision and then eventually to machine learning. The combination of Swift and machine learning lured me to the incredible <a href="https://www.tensorflow.org/swift/guide/overview">Swift for TensorFlow</a> team at Google, and then here to <a href="https://passivelogic.com/">PassiveLogic</a>.</p><p>Unfortunately, I got so busy with all the new things I was working on that I had less and less time to maintain Molecules over the years, and it eventually fell into disrepair. That had eaten at me, so I recently started rewriting it from the original Objective-C, UIKit, and OpenGL ES to modern Swift, SwiftUI, and Metal. Going through that process is showing me both how far technology has advanced in the last 15 years as well as how much I’ve grown as a developer. I‘m having a lot of fun, and hope to have it back on the App Store soon.</p><figure><img alt="My pug Bucky is sleeping next to me while I work on a SwiftUI interface preview on my laptop." src="https://cdn-images-1.medium.com/max/1024/1*vuJsbHwpT-Y4y-ZaIP2PxA.jpeg" /><figcaption>My pug Bucky is helping me get my Metal rendering correct in my SwiftUI interface.</figcaption></figure><p>I truly am grateful for the opportunity I had to be there at the start of something special 15 years ago. I have a chance to work with outstanding people on fascinating technology every day at PassiveLogic in large part because of what came out of my little open source project. To say nothing of the wonderful wife I might have never met without this.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MbJThqI1rqp-Llj-TLHW5Q.png" /></figure><p>Even though my focus at PassiveLogic is on our deep technology stack, primarily <a href="https://medium.com/passivelogic/recent-advances-in-differentiable-swift-f60ccc0c6f7e">how we can advance the Swift programming language</a> to support generative autonomy, I also love working with our top-notch iOS development team. We’ll have something really exciting to show off soon that combines 3D rendering with the latest in spatial computing.</p><p>It’s thrilling to be there at the launch of a new platform, and I’m hoping that the one we’re building here will give my younger coworkers the same sense of pride and awe that I had 15 years ago.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f0acd07e44a7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/15-years-of-the-iphone-app-store-the-story-of-molecules-f0acd07e44a7">15 Years of the iPhone App Store: The Story of Molecules</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Debugging the Swift Compiler, Part 1: Writing Good Bug Reports]]></title>
            <link>https://medium.com/passivelogic/debugging-the-swift-compiler-part-1-writing-good-bug-reports-8357cfc459e?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/8357cfc459e</guid>
            <category><![CDATA[pltech]]></category>
            <category><![CDATA[swift]]></category>
            <category><![CDATA[compilers]]></category>
            <category><![CDATA[automatic-differentiation]]></category>
            <category><![CDATA[swift-programming]]></category>
            <dc:creator><![CDATA[Brad Larson]]></dc:creator>
            <pubDate>Fri, 14 Jul 2023 21:03:32 GMT</pubDate>
            <atom:updated>2023-07-14T23:48:30.448Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*V-W9uJc5N2AWUR0RQCtWyw.png" /></figure><p>For the vast majority of programmers, when our code doesn’t build or doesn’t run correctly it’s almost always our fault. The compiler is pretty much the last thing we blame. However, at PassiveLogic, we push the limits of the Swift programming language, in particular by using <a href="https://medium.com/passivelogic/intro-to-differentiable-swift-part-0-why-automatic-differentiation-is-awesome-a522128ca9e3">an experimental language feature</a> and <a href="https://medium.com/passivelogic/recent-advances-in-differentiable-swift-f60ccc0c6f7e">helping in its evolution</a>. As a result, “my code broke the compiler” is something you hear a little more frequently around here.</p><p>One of the awesome things about the Swift language is that its compiler and other components <a href="https://github.com/apple/swift">are open source</a>, and anyone can report and even fix issues that they encounter. We’re investing in a team to do just that in support of PassiveLogic’s <a href="https://passivelogic.com/">ecosystem of products</a> that use Swift under the hood. This article is the first in a planned series about the process we use when encountering a bug in the current implementation of the Swift language and how we approach isolating, reporting, and fixing that issue.</p><p>Before you can fix a bug, you need to first identify it. Writing a great bug report is an undervalued skill, and can be the difference between a bug sitting there for years or it being fixed in days. The best bug reports make it immediately obvious what’s going wrong, where the problem may lie, and how to verify a fix. Filing an easy-to-reproduce test case for a bug, with a detailed description of how something is deviating from expected behavior, can often be half the battle:</p><figure><img alt="A pie chart, in G.I.Joe’s color scheme, with Knowing at 50% and Red Lasers and Blue Lasers at 25% each." src="https://cdn-images-1.medium.com/max/713/1*rz2m0XUHDdR59s5RX18Nkg.jpeg" /><figcaption><a href="http://nerduo.com/">nerduo.com</a></figcaption></figure><p>Compiler bugs come in a variety of shapes, but one that Swift developers may have encountered (especially if you worked with the language in its early days or have tried out new features) is the compiler crasher. You try to build some code and the compilation process bombs out with a stack trace at the command line or Xcode gives you a cryptic</p><pre>Command SwiftCompile failed with a nonzero error code</pre><p>Let’s say you’ve hit a genuine compiler crasher like this in your code and you want to report it. Before you do so, it’s entirely possible that this bug has already been fixed upstream in the compiler. If you have a project that allows it, try <a href="https://www.swift.org/download/">downloading a newer Swift toolchain from Swift.org</a> with the very latest being one of the (nearly) nightly trunk snapshots. Build your code again using that newer toolchain and verify that it still repeats or at least causes a new crasher with that.</p><p>If the failure goes away, that’s a strong indication that someone has fixed the problem in the meantime. Fixes and improvements in the main branch of Swift usually appear first in nightly toolchain snapshots on Swift.org, then release branch candidates on Swift.org, and finally the toolchains that ship with versions of Xcode. (We’ll talk in a future article about how to build your own Swift toolchain from scratch if you want the very latest code or want to customize the toolchain for your own needs.)</p><p>Assuming the crasher still appears in the latest Swift toolchains, it might be a good idea to check if others have reported it. Before doing that, let’s see what information is provided to us in a stack trace from a typical compiler crasher. This will all be printed to the console when building at the command line or in most build tools, but if you build in Xcode and you only see the above one-line error message, go to the build log and click the detail disclosure icon to show the full stack trace:</p><figure><img alt="An image of the location of the detail disclosure box for build failures, at the far right of the compile error." src="https://cdn-images-1.medium.com/max/812/1*r1f7IMiRUJEJ-80ywDQjPw.png" /></figure><p>Many compiler crashers are due to code leading to unexpected conditions and causing assertion failures. The Swift toolchains that ship with Xcode have many assertions turned off, but the Swift.org toolchains have them turned on. By building with <a href="http://swift.org/">Swift.org</a> toolchains, you may encounter an assertion failure closer to what’s actually going wrong inside the compiler, so that’s another good reason to try your code against one of the Swift.org toolchains.</p><p>If you start reading down the stack trace, one of the first things you will encounter is the assertion failure itself, which may look like the following:</p><pre>Assertion failed: (hasVal), function value, file Optional.h, line 216.</pre><p>That’s your first clue, and there might be sufficiently distinct wording in the assertion failure to start searching for previously reported issues. Currently, all Swift compiler issues are <a href="https://github.com/apple/swift/issues">hosted on GitHub</a> and you can try searching the open issues for the specific assertion failure you’re seeing or for aspects of your code that you think might be related. If someone has reported your exact assertion failure and has reproducible code that is similar to what’s failing locally, great. In our case, we usually don’t find our assertion failures covered in the open issues and we go on to dig a bit deeper.</p><p>After the assertion failure message, there’s usually some more interesting information in a stack dump:</p><pre>1. Apple Swift version 5.9-dev (LLVM f58e4c448dc4009, Swift 07f938435a8b7d1)<br>2. Compiling with the current language version<br>3. While evaluating request ExecuteSILPipelineRequest(Run pipelines { Mandatory Diagnostic Passes + Enabling Optimization Passes } on SIL for KeyPathMapCrasher)<br>4. While running pass #382 SILModuleTransform &quot;Differentiation&quot;.<br>5. While canonicalizing `differentiable_function` SIL node   %25 = differentiable_function [parameters 0 1] [results 0] %24 : $@callee_guaranteed (SomeDifferentiableStruct, Double) -&gt; Double // users: %52, %26<br>6. While ...in SIL function &quot;@$s17KeyPathMapCrasher12someFunctionyyKF&quot;.<br> for &#39;someFunction()&#39; (at /Users/larson/Developer/DifferentiableSwift/CompilerBugs/KeyPathMapCrasher/KeyPathMapCrasher/main.swift:34:8)<br>7. While processing // differentiability witness for setAndWrite #1 (aStruct:newValue:) in someFunction()<br>sil_differentiability_witness private [reverse] [parameters 0 1] [results 0] @$s17KeyPathMapCrasher12someFunctionyyKF11setAndWriteL_7aStruct8newValueSdAA018SomeDifferentiableJ0V_SdtF : $@convention(thin) (SomeDifferentiableStruct, Double, @guaranteed { var Cacher&lt;SomeDifferentiableStruct, Double&gt; }) -&gt; Double {<br>}<br><br> on SIL function &quot;@$s17KeyPathMapCrasher12someFunctionyyKF11setAndWriteL_7aStruct8newValueSdAA018SomeDifferentiableJ0V_SdtF&quot;.<br> for &#39;setAndWrite(aStruct:newValue:)&#39; (at /Users/larson/Developer/DifferentiableSwift/CompilerBugs/KeyPathMapCrasher/KeyPathMapCrasher/main.swift:37:5)<br>8. While generating VJP for SIL function &quot;@$s17KeyPathMapCrasher12someFunctionyyKF11setAndWriteL_7aStruct8newValueSdAA018SomeDifferentiableJ0V_SdtF&quot;.<br> for &#39;setAndWrite(aStruct:newValue:)&#39; (at /Users/larson/Developer/DifferentiableSwift/CompilerBugs/KeyPathMapCrasher/KeyPathMapCrasher/main.swift:37:5)<br>9. While generating pullback for SIL function &quot;@$s17KeyPathMapCrasher12someFunctionyyKF11setAndWriteL_7aStruct8newValueSdAA018SomeDifferentiableJ0V_SdtF&quot;.<br> for &#39;setAndWrite(aStruct:newValue:)&#39; (at /Users/larson/Developer/DifferentiableSwift/CompilerBugs/KeyPathMapCrasher/KeyPathMapCrasher/main.swift:37:5)</pre><p>and a stack dump without symbols:</p><pre>0  swift-frontend           0x000000010575c184 llvm::sys::PrintStackTrace(llvm::raw_ostream&amp;, int) + 56<br>1  swift-frontend           0x000000010575b528 llvm::sys::RunSignalHandlers() + 112<br>2  swift-frontend           0x000000010575c7c4 SignalHandler(int) + 304<br>3  libsystem_platform.dylib 0x000000019a96ea24 _sigtramp + 56<br>4  libsystem_pthread.dylib  0x000000019a93fc28 pthread_kill + 288<br>5  libsystem_c.dylib        0x000000019a84dae8 abort + 180<br>6  libsystem_c.dylib        0x000000019a84ce44 err + 0<br>7  swift-frontend           0x00000001058a6a30 swift::autodiff::PullbackCloner::Implementation::getAdjointBuffer(swift::SILBasicBlock*, swift::SILValue) (.cold.1) + 0<br>8  swift-frontend           0x000000010103b428 swift::autodiff::PullbackCloner::Implementation::getTangentValueCategory(swift::SILValue) + 260<br>9  swift-frontend           0x00000001010389a4 swift::autodiff::PullbackCloner::Implementation::run() + 2280<br>10 swift-frontend           0x0000000101038084 swift::autodiff::PullbackCloner::run() + 24<br>11 swift-frontend           0x0000000101054e78 swift::autodiff::VJPCloner::Implementation::run() + 1368<br>12 swift-frontend           0x0000000101055540 swift::autodiff::VJPCloner::run() + 24<br>13 swift-frontend           0x000000010118cb00 (anonymous namespace)::DifferentiationTransformer::canonicalizeDifferentiabilityWitness(swift::SILDifferentiabilityWitness*, swift::autodiff::DifferentiationInvoker, swift::IsSerialized_t) + 5616<br>14 swift-frontend           0x000000010118f674 (anonymous namespace)::DifferentiationTransformer::promoteToDifferentiableFunction(swift::DifferentiableFunctionInst*, swift::SILBuilder&amp;, swift::SILLocation, swift::autodiff::DifferentiationInvoker) + 3924<br>15 swift-frontend           0x000000010118cd50 (anonymous namespace)::DifferentiationTransformer::processDifferentiableFunctionInst(swift::DifferentiableFunctionInst*) + 468<br>16 swift-frontend           0x000000010118ad2c (anonymous namespace)::Differentiation::run() + 1380<br>17 swift-frontend           0x0000000101217968 swift::SILPassManager::runModulePass(unsigned int) + 980<br>18 swift-frontend           0x000000010121d964 swift::SILPassManager::execute() + 624<br>19 swift-frontend           0x00000001012148fc swift::SILPassManager::executePassPipelinePlan(swift::SILPassPipelinePlan const&amp;) + 68<br>20 swift-frontend           0x0000000101214880 swift::ExecuteSILPipelineRequest::evaluate(swift::Evaluator&amp;, swift::SILPipelineExecutionDescriptor) const + 68<br>21 swift-frontend           0x0000000101253d78 swift::SimpleRequest&lt;swift::ExecuteSILPipelineRequest, std::__1::tuple&lt;&gt; (swift::SILPipelineExecutionDescriptor), (swift::RequestFlags)1&gt;::evaluateRequest(swift::ExecuteSILPipelineRequest const&amp;, swift::Evaluator&amp;) + 28<br>22 swift-frontend           0x000000010123d16c llvm::Expected&lt;swift::ExecuteSILPipelineRequest::OutputType&gt; swift::Evaluator::getResultUncached&lt;swift::ExecuteSILPipelineRequest&gt;(swift::ExecuteSILPipelineRequest const&amp;) + 252<br>23 swift-frontend           0x0000000101214af0 swift::executePassPipelinePlan(swift::SILModule*, swift::SILPassPipelinePlan const&amp;, bool, swift::irgen::IRGenModule*) + 84<br>24 swift-frontend           0x000000010123f210 swift::runSILDiagnosticPasses(swift::SILModule&amp;) + 192<br>25 swift-frontend           0x0000000100b399a4 swift::CompilerInstance::performSILProcessing(swift::SILModule*) + 68<br>26 swift-frontend           0x0000000100997ec0 performCompileStepsPostSILGen(swift::CompilerInstance&amp;, std::__1::unique_ptr&lt;swift::SILModule, std::__1::default_delete&lt;swift::SILModule&gt;&gt;, llvm::PointerUnion&lt;swift::ModuleDecl*, swift::SourceFile*&gt;, swift::PrimarySpecificPaths const&amp;, int&amp;, swift::FrontendObserver*) + 792<br>27 swift-frontend           0x00000001009975b0 swift::performCompileStepsPostSema(swift::CompilerInstance&amp;, int&amp;, swift::FrontendObserver*) + 600<br>28 swift-frontend           0x00000001009a6944 withSemanticAnalysis(swift::CompilerInstance&amp;, swift::FrontendObserver*, llvm::function_ref&lt;bool (swift::CompilerInstance&amp;)&gt;, bool) + 160<br>29 swift-frontend           0x000000010099a128 performCompile(swift::CompilerInstance&amp;, int&amp;, swift::FrontendObserver*) + 740<br>30 swift-frontend           0x0000000100999098 swift::performFrontend(llvm::ArrayRef&lt;char const*&gt;, char const*, void*, swift::FrontendObserver*) + 2504<br>31 swift-frontend           0x00000001007ec354 swift::mainEntry(int, char const**) + 2144</pre><p>These have a lot of useful information in them that can help identify where, exactly, the compiler tripped and fell down. The first stack trace tells us in item 6 that the crash occurred when processing someFunction() at line 34, column 8 of main.swift. That already shows where to start looking in your code.</p><p>Within the second stack trace, lines 0-6 can be stepped over because those are all for the supporting code to print the stack trace. The good stuff starts from line 7 down. We can see that <a href="https://github.com/apple/swift/blob/b19fc60a529a94c8297fb39e16e712e5ea58acb2/lib/SILOptimizer/Differentiation/PullbackCloner.cpp#L556">getAdjointBuffer() in </a><a href="https://github.com/apple/swift/blob/b19fc60a529a94c8297fb39e16e712e5ea58acb2/lib/SILOptimizer/Differentiation/PullbackCloner.cpp#L556">PullbackCloner.cpp</a>is where the compiler encountered the assertion failure. We can also see the entire stack trace of calls in the compiler that led to that. From the perspective of a compiler engineer, the value of this information is apparent and thus hugely helpful in a bug report.</p><p>We’ve now learned a bit about what could be going wrong and where, and we can start filing a bug report based on the above, but the very best reports also include easy-to-use steps for reproducing the problem. In the case of a compiler crasher, the reproducer details include the bare minimum code that brings out the crash, ideally within a single file that has no other dependencies.</p><p>The best reproducer cases for Swift compiler bugs are single, self-contained files that can be built directly using the command swiftc test.swift (or swiftc -O test.swift, if this bug only occurs in optimized builds) and have the absolute minimum amount of code needed to trigger the issue. (Please make sure to remove or anonymize all references to proprietary code in your reproducers before posting them in public.)</p><p>A single file is preferred because it is generally easier to debug the compiler running against one file than if you need to use the full Swift Package Manager infrastructure to build a project.</p><p>When minimizing a reproducer from a large codebase, I always start with what the compiler stack trace is telling me. In the above example, the stack trace narrows down the failure to the line of code that causes the problem. Take a look at the code around that point. Developers around the world use Swift every day without it crashing, so something about your code has to be different enough to cause the compiler to break. Look for interesting combinations of features (in our case, differentiable Swift’s special valueWithGradient(at:of) function and throwing functions).</p><p>Once you have a general location and what you think might be the items leading to the failure, start ablating off code to narrowly isolate the failure point. Ideally, this will be a quick process and you can pull out a single standalone function and have it still break the compiler. Unfortunately, sometimes it can be a much slower process of isolation. Remove things until the compiler stops breaking, then roll back until it does again and iterate until you’ve sufficiently isolated the problem area.</p><p>My coworker Martin is especially talented at tracking down reproducers for Swift compiler bugs and writing up reports on them. His <a href="https://github.com/apple/swift/issues/66312">bug reports</a> have led to <a href="https://github.com/apple/swift/pull/66800">quick fixes</a> for some longstanding Swift compiler issues. I asked him for some advice on how he approaches minimizing a compiler crasher from a large codebase:</p><blockquote>One of the first goals in generating a reproducer from a large codebase is to reduce build times. Because there is a significant amount of trial and error involved in determining which dozen or two lines of code cause the compiler to crash, speeding up each iteration is high on the list of priorities.</blockquote><blockquote>This process can include removal of entire unit test suites which have nothing to do with the offending code, or preventing the compilation of entire modules that are not involved with the code in question.</blockquote><blockquote>At this point, it’s useful to use a small utility that simply replaces each non-differentiable Swift function body with a call to fatalError(). Once a given file has had its function bodies replaced as such, it will not only further reduce build times, but also highlight code which was not contributing to the crasher in the first place. Once large swaths of the codebase have been processed in this way, we can then begin removing entire structs or classes that are no longer referenced.</blockquote><p>Hopefully, at the end of this process, you’ll have a small reproducer that still breaks the latest nightly builds of the compiler. Now it’s time to actually write up your bug report. Head on over to <a href="https://github.com/apple/swift/issues">apple/swift’s GitHub issues</a> and create a new issue with a descriptive title. Try to be more descriptive than “compiler crashes with this code” and include some unique details (assertion failure, where the crash happens, etc.). In the body of the bug report, list out the assertion failure, all stack traces, and describe your reproducer such that anyone can copy it and run it locally.</p><p>As an excellent example of the kind of bug report that can result from the process described above, I direct you to <a href="https://github.com/apple/swift/issues/63331">apple/swift GitHub issue #63331</a> that Martin filed in January. The assertion failure and stack traces I used as examples above were drawn from that issue, and you can see how he clearly lays out each piece of information and provides a minimal, single-file reproducer case that’s easy to run. Armed with this kind of information, compiler engineers are well-prepared to start diagnosing and then fixing the crasher — that will be the subject of our next posts in the series.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8357cfc459e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/debugging-the-swift-compiler-part-1-writing-good-bug-reports-8357cfc459e">Debugging the Swift Compiler, Part 1: Writing Good Bug Reports</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Celebrating International Women in Engineering Day at PassiveLogic]]></title>
            <link>https://medium.com/passivelogic/celebrating-international-women-in-engineering-day-at-passivelogic-2cb6277f83e0?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/2cb6277f83e0</guid>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Caroline Genster]]></dc:creator>
            <pubDate>Fri, 23 Jun 2023 19:08:12 GMT</pubDate>
            <atom:updated>2023-06-23T19:08:12.555Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zhgi-xJtOwruVJPeNA2vZg.png" /></figure><p>International Women in Engineering Day honors women in engineering roles around the world, highlighting their contributions, creativity, and leadership that drive their industries forward. The need to cultivate an interest in engineering among women, and support those women is acute. According to the <a href="https://www.aauw.org/resources/research/the-stem-gap/">American Association of University Women</a>, women make up only 28 percent of the work force in science technology, engineering, and math. Further, a recent study by the <a href="https://www.iea.org/topics/energy-and-gender">International Energy Agency</a> (IEA) concluded that women only account for 16 percent of the traditional energy sector. As a company sitting at the intersection of cutting edge engineering and the future of energy management and control, PassiveLogic recognizes the urgency of building and retaining a diverse workforce that supports the best engineers in the world, who ever and where ever they are.</p><p>Simply <em>wanting</em> to foster equity and diversity is not enough. PassiveLogic’s leadership designed the company to support equality in several critical ways, including its organizational design and compensation calculation.</p><p>All employees are only 1–2 steps away from the CEO, enabling all employees to collaborate directly with leadership, be encouraged to innovate at all stages, especially as early-career employees. As a result, 5 out of 13 of our technical leads started off at PassiveLogic as entry-level engineers, and of those 5 technical leads who rose to their position from entry-level, two of them are women. Today, these incredible women run their own technical teams that contribute directly to the company’s innovation.</p><p>PassiveLogic also developed its Equitable Compensation Calculator as a tool to operationalize our core principles of equity and sustainability. We generate employment offers and salary increase using the Equitable Compensation Calculator. It considers objective inputs like experience, expertise, education, technical challenge score (if applicable), and location-based cost of living as the only parameters to ensure equitable compensation across gender, age, ethnicity, employment gaps, and other identity factors. This ensures variations in employee pay are accounted for based on merit. Further, it supports PassiveLogic’s core value of employee ownership (ensuring equity and an ethos of “we all win together”) by calculating Employee Stock Options as part of each employee’s total compensation.</p><p>The IEA report further noted: “There is an urgency for countries to attract and retain a diverse workforce in the energy sector to ensure innovation and the inclusive perspectives needed to successfully navigate the low‑carbon energy transition.” We decided to talk with some of the engineers who are women at PassiveLogic about their careers, their perspectives, and what it means to them to work at company helping lead the energy transition.</p><p>— —</p><p>“I like the Earth and I like to be challenged,” remarked Katerina Tafoya, a Tools Software Engineer on the Cloud Team at PassiveLogic, when asked why she joined the company. Katerina is not alone in that sentiment. Across the board, the engineers at PassiveLogic appreciated the unique convergence of opportunities available to them here. Alexa Kalandiak, Head of Physical Product Design, said she valued working on a product that will have a net good on the world. “I’m not creating just to put something on an Amazon shelf.”</p><p>Before her time as a software engineer, Sarah Schuster-Johnson, Quantum Lead Software Engineer, was a documentary filmmaker focused on telling stories that could help change the world. When she went back to school for a Masters in Software Development, she was driven by the same passion. She joined the company in her final semester of school as an intern, attracted by its focus on the environment and sustainability combined with its start-up stage. “Having a larger impact on our carbon footprint is really attractive to me,” she remarked.</p><p>Zohreh Shaghaghian, a Generative Design Engineer on the Building Studio Team, also came to PassiveLogic through a less traditional route. She holds a bachelors, masters, and PhD, all in architecture! Throughout her education, Zohreh explored the intersection of digital innovation and architecture, working to understand how architecture could adopt more modern technologies to enable better design and collaboration with all building stakeholders. That brought her to PassiveLogic, where she provides her team with unique insights on what the future users of her team’s product will need and want.</p><p>Unanimous among our engineers was the sentiment that PassiveLogic has challenged them intellectually and supported their growth. Katerina noted that at PassiveLogic there are constant opportunities for learning. Alexa added that in her time here, she has worked with the smartest people she’s ever met. Sarah appreciated that all the work at PassiveLogic combines team-work with a sense of ownership over your work. “You’re not just a cog in the machine.” Zohreh remarked, “The people I’m working with are talented, supportive, and open-minded in training. They let you explore and grow and don’t expect you to come with a static background.”</p><p>When asked what advice they would give to younger women interested in engineering, our engineers all emphasized the importance of confidence, especially in addressing feelings of imposter syndrome. Alexa remarked, “if you want to be here, you belong here. You have a right to be here as much as anyone else.” Our team members emphasized that engineers from non-traditional backgrounds offer unique insights on their work that may not have been considered before. Katerina said, “I think differently as a woman. When I chime in, I have different perspectives.” Sarah noted that it is critical that those with diverse backgrounds in engineering speak up: “Don’t be timid. Your background means you bring a different perspective that needs to be heard.” And when faced with new challenges, Zohreh, Katerina, and Alexa all highlighted the importance of asking questions. Alexa continued, “Ask questions. And then ask more questions. It’s accelerated my learning.” Katerina said, “Don’t give up. It’s going to be hard, but anything worth doing will be hard.” Zohreh echoed Katerina saying, “Don’t be afraid. It’s okay to step into an area you don’t know well. Step into that area and have the confidence in yourself that you can learn it.”</p><p>Building a team comprised of diverse perspectives enables PassiveLogic to build the ground-breaking, innovative technology stack that unlocks and democratizes the future of generative autonomy.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2cb6277f83e0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/celebrating-international-women-in-engineering-day-at-passivelogic-2cb6277f83e0">Celebrating International Women in Engineering Day at PassiveLogic</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What it takes to #EmbraceEquity in the energy industry]]></title>
            <link>https://medium.com/passivelogic/what-it-takes-to-embraceequity-in-the-energy-industry-a2a415dcccd9?source=rss----b3718cadbd32---4</link>
            <guid isPermaLink="false">https://medium.com/p/a2a415dcccd9</guid>
            <category><![CDATA[international-womens-day]]></category>
            <category><![CDATA[equity]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[women-in-tech]]></category>
            <category><![CDATA[autonomous-building]]></category>
            <dc:creator><![CDATA[Travis Anna Harvey]]></dc:creator>
            <pubDate>Wed, 08 Mar 2023 00:50:39 GMT</pubDate>
            <atom:updated>2023-03-08T00:50:39.824Z</atom:updated>
            <content:encoded><![CDATA[<p><em>When companies implement equitable practices, it’s a win for all</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tVlC4Uj-fCEFI7Q2n3RIdw.jpeg" /><figcaption>PassiveLogic at the Women Tech Council 2022 Women Tech Awards</figcaption></figure><p>By Travis Anna Hallstrom, Head of Brand and Product Strategy at PassiveLogic</p><p>As autonomous buildings technologists, our industry is part of the broader energy industry. Buildings consume 41% of the world’s energy resources, and autonomous buildings reduce that consumption by on average 30% — huge impact on that consumption. I’ve been engaged in a deep dive into the energy sector’s labor market and a few important items stand out.</p><p>According to <a href="https://www.iea.org/topics/energy-and-gender?utm_campaign=ENT-03MAR23-North%20America/LATAM&amp;utm_medium=email&amp;utm_source=Eloqua">the IEA</a>, women account for only 22% of the traditional energy sector workforce. For management levels, the numbers are lower. Even more alarming, wages for female employees within this sector are almost 20% lower than their male counterparts, and this gap persists even when factors like skills, education, and experience are normalized — indicating that the gap may have more to do with gender than it does expertise (<a href="https://www.iea.org/articles/understanding-gender-gaps-in-wages-employment-and-career-trajectories-in-the-energy-sector">OECD LinkEED project</a>).</p><blockquote>Our mission is clear: implement practices for equitable action to launch our industry forward.</blockquote><p>These findings communicate a clear message, articulated by the International Energy Agency in their latest report on gender and energy: “There is an urgency for countries to attract and retain a diverse workforce in the energy sector to ensure innovation and the inclusive perspectives needed to successfully navigate the low‑carbon energy transition.” In other words, gender equity is not just a “nice to have,” adding to the endless list of trending workplace priorities crowd-sourced by folks not in the trenches — but is an urgent “must-have,” necessary to achieve our ambitious and necessary goals for the industry. We must incorporate more equitable practices in our workplaces to fully leverage the skills, expertise, and innovations from the <em>entire</em> workforce.</p><p><strong>Equity in the Energy Sector</strong></p><p>This year’s theme for <a href="https://www.internationalwomensday.com/">International Women’s Day</a> (March 8) is #EmbraceEquity. The aim of this theme is to highlight why equal <em>opportunities</em> are not enough for sustained change, but we must commit to equitable <em>action</em> to foster the inclusion and belonging that will define the global change we seek.</p><p>What does equitable action really mean? In <a href="https://www.unfpa.org/resources/frequently-asked-questions-about-gender-equality">the context of gender</a>, equity is the practice of being fair to women and men so that they have a level playing field with equal access to opportunities, taking into account the different starting point of different identity groups. To create this level playing field, it is necessary to implement practices, strategies, and policies to both remove negative bias based on identity, and correct with positive bias. Our objective is that by practicing equitable action, equity ultimately leads to equality across gender.</p><p><strong>Our mission is clear: implement practices for equitable action to launch our industry forward.</strong> One of the more striking findings of the <a href="https://www.iea.org/articles/understanding-gender-gaps-in-wages-employment-and-career-trajectories-in-the-energy-sector">OECD LinkEED project</a> is women are more likely than men to leave the energy sector altogether. This diaspora is a lagging indicator of the unequal distribution by gender in lower-paying (non-leadership) roles and lower-paying firms. These combined factors exacerbate the gender wage gap as men advance in their careers within the sector, while women stay behind in more junior roles. Also from the IEA report: “Men advance in their careers by moving to more productive and high-wage firms within the sector, while women tend to stay behind, increasing the gender wage gap between firms with age.” The entire industry suffers as women — who make up 40% of the talent pool — leave the energy sector in droves.</p><p>This finding highlights the importance of retaining the women that already make up our workforce in the energy (and building automation) industry. At PassiveLogic our practice of nurturing existing talent has borne meaningful fruit. Our tailored mentorships and flat organizational structure give early-career employees an opportunity to work closely with executives who encourage them to develop their skills. We have a number of day-to-day operational practices that encourage meritocratic review of new ideas, and encourage contributions of junior team members. As a result, 5 out of 13 of our technical leads started off at PassiveLogic as entry-level engineers, and of those 5 technical leads who rose to their position from entry-level, two of them are women. Today, these incredible women run their own technical teams that contribute directly to the company’s innovation.</p><p><strong>How Equity Transforms Businesses</strong></p><p>As we deliberately and consistently incorporate practices to create a more equitable workplace, we cannot ignore that pay equity is essential to leveling the playing field. At PassiveLogic, our Equitable Salary Calculator was developed as a tool to operationalize our core principles of equity and sustainability. We generate every employment offer and salary increase using the Equitable Salary Calculator. It considers objective inputs like experience, expertise, education, technical challenge score (if applicable), and location-based cost of living as the only parameters to ensure equitable compensation across gender, age, ethnicity, employment gaps, and other identity factors. This ensures variations in employeepay are accounted for solely based on merit.</p><p>This equitable pay practice has benefitted the company in many ways. Last year, we nearly doubled the number of women in our company, including doubling the number of women in technical roles. We were pleased to announce that we increased the number of women in leadership at our company from 1 to 4, making up a quarter of our greater leadership team, creating additional opportunities for female mentorship. It is no understatement that we mean it when we say we want to give opportunities to women. I am proud of the growth we have been able to achieve by recruiting more women and mentoring our female employees so they can grow in their careers, and contribute to the company’s success.</p><p><strong>Celebrating the Wins</strong></p><p>Adopting a holistic view of the journey to #EmbraceEquity involves celebrating our progress and recognizing the benefits that equity can bring to the industry.</p><p>Diverse teams are 70% more likely to capture new markets, and are 120% more likely to hit financial goals <a href="https://www.fundera.com/resources/diversity-in-the-workplace-statistics">(Fundara by NerdWallet)</a>. It is clear that diversity isn’t just necessary to build a psychologically safe and equitable environment for employees, but can provide an array of opportunities for the company itself.</p><p>Last year, I was humbled to receive the Transformation Leadership Award during the 2022 Women Tech Awards. This award is in an indicator that PassiveLogic’s efforts towards building an equitable workforce are valued by the larger tech community. Looking back on this past year where we have experienced a growth from 61 to 100+ employees, we can attribute many of our technological achievements to our increasingly diverse workforce that provides us new perspectives and a greater pool of knowledge from which to draw.</p><p>At PassiveLogic, we’re passionate about data-driven continuous improvement practices. This approach has led to creating inclusive pathways for women and diverse candidates, which has in turn prepared us to more effectively achieve a low-carbon energy transition.</p><p><em>If you’re interested in disrupting the world of automation with the first fully autonomous platform for buildings, check out our job openings here: </em><a href="https://passivelogic.com/careers/"><em>https://passivelogic.com/careers/</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a2a415dcccd9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/passivelogic/what-it-takes-to-embraceequity-in-the-energy-industry-a2a415dcccd9">What it takes to #EmbraceEquity in the energy industry</a> was originally published in <a href="https://medium.com/passivelogic">PassiveLogic</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>