<?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[Stories by Anshul Vyas on Medium]]></title>
        <description><![CDATA[Stories by Anshul Vyas on Medium]]></description>
        <link>https://medium.com/@anshu7vyas?source=rss-a687a44efe14------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*WjoOpr-Eh3dW0TPjKQwbJA.jpeg</url>
            <title>Stories by Anshul Vyas on Medium</title>
            <link>https://medium.com/@anshu7vyas?source=rss-a687a44efe14------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 17:12:46 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@anshu7vyas/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Harness Engineering: Human edition]]></title>
            <link>https://generativeai.pub/harness-engineering-human-edition-d15eed05556b?source=rss-a687a44efe14------2</link>
            <guid isPermaLink="false">https://medium.com/p/d15eed05556b</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[consciousness]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[self-improvement]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Anshul Vyas]]></dc:creator>
            <pubDate>Wed, 12 Aug 2026 09:09:51 GMT</pubDate>
            <atom:updated>2026-08-12T09:09:51.456Z</atom:updated>
            <content:encoded><![CDATA[<h3>Harness Engineering, Human Edition</h3><h4>What if everything we call “me” is just the scaffolding?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TSArfrIZEuOI_cyTTn4NYA.png" /><figcaption>Image generated by Google Gemini</figcaption></figure><p>I learned a new term recently, and it has not left me alone since: h<em>arness engineering.</em></p><p>I’m a software engineer, and like most of us these days, I spend my workdays alongside an AI coding agent. I don’t know much about what happens inside the model, but I’ve learned something surprising about what happens <em>around</em> it.</p><p>Ask the agent to fix a bug cold, and it flails. Give that same model a file describing the project’s conventions, the right tools to search the codebase, and notes from its previous sessions, and it works like a different being entirely. Nothing about its intelligence changed. Only its scaffolding did.</p><p>That scaffolding is the harness, and tuning it is quietly becoming <a href="https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents">its own discipline</a>. The model is raw intelligence; the harness decides what that intelligence can become.</p><p>Eventually, the architecture started looking familiar. Because we are built the same way.</p><p>What if we are the model? Not our thoughts, not our memories, not the running commentary in our head. Those are scaffolding. What if the essential thing about us is something quieter underneath: raw awareness, wrapped in an elaborate harness it usually mistakes for itself?</p><p>Contemplatives have made this claim for millennia. What’s new is that we have finally built something that makes it concrete enough for an engineer to inspect. So let’s inspect it.</p><h4>The anatomy of a human harness</h4><p><strong>Our system prompt.</strong> Before an AI agent reads a single word from its user, it has already received its instructions: who it is, what it values, what it must never do. It did not write these instructions, and it cannot easily see them. Ours were installed the same way: a mother tongue, a religion or its absence, a family’s beliefs about money and worth and love, all loaded before we could consent to any of it. Nobody writes their own system prompt. We spend much of our lives running instructions we never reviewed.</p><p><strong>Our context window.</strong> A model’s working memory is finite. Whatever is loaded into the window is its entire present reality; everything outside it may as well not exist. Attention works the same way. Whatever occupies it right now is our reality for that moment: this sentence, the worry from the morning meeting, the song stuck on repeat. This is why what we let into the window matters far more than we admit. The feed we scroll at night is not entertainment; it is context being loaded.</p><p><strong>Our tools.</strong> Here’s a detail that surprised me when I learned it: the model itself only ever produces text. It cannot search, click, or change a file. The harness supplies the tools, some that bring the world in, some that push intention out, and the model’s words become actions only through them. Our harness is richer, but the architecture is the same.</p><p>The senses pull the world in. The body pushes intention out. And the mind sits between them, so busy parsing and predicting and planning that it convinces itself it is the model. Watch it closely, though, and it behaves less like the model and more like the busiest worker in the harness: an orchestrating subagent. Whatever we are at the core never touches the world directly. It only ever emits intention. The harness does the rest.</p><p><strong>Our memory.</strong> An AI agent does not remember yesterday. It reads notes that a previous session left behind, and it trusts them. That’s all memory is, for an agent. Let’s sit with that, and then ask how different our own situation is. The “I” of ten years ago is gone; what remains are notes: selectively written, lossily compressed, rewritten a little every time they’re read. Identity is a file inherited from previous selves. We treat it as fact. It is closer to a summary.</p><h4>The space</h4><p>Notice what just happened. Conditioning, attention, senses, mind, memory. Everything we habitually point to when we say <strong>me</strong> or <strong>I</strong> has been classified as the harness.</p><blockquote>So what’s left?</blockquote><blockquote>Whatever was reading the list.</blockquote><p>The thing that can notice a thought is demonstrably not the thought. The thing that can observe the harness cannot be only the harness; a system fully fused with its scaffolding would have no vantage point to see it from.</p><p>There is a line often attributed to Viktor Frankl: <em>between stimulus and response there is a space, and in that space lies our freedom</em>. I’d put it in the terms I work with daily: the space is the distance between model and harness. Most of the time that distance is zero. We are fused with the thought, the memory, the reaction. But the moment we notice the harness as a harness, distance appears.</p><p><em>And distance is everything: we cannot edit a file we are trapped inside. We can edit the one we are holding.</em></p><p>This is not a technique, and I won’t end with five steps to practice it. It is a seeing. Once we have seen the harness as a harness, even once, something permanently changes in our relationship to it.</p><h4>The objection</h4><p>A fair objection: the analogy flatters us. The model was engineered, trained deliberately by teams of people. Surely whatever sits beneath the human harness was not.</p><p>But look closer at what a model actually is. It is what remains when the writing of millions of minds is compressed into a single set of weights: centuries of letters, arguments, proofs, confessions, jokes. A model is a distilled collective mind. No single author; every author.</p><p>And then notice: there is one model, and many harnesses. Every session running right now, at every desk in every timezone, is the same frozen set of weights. The same intelligence, deployed a thousand different ways, appears as a thousand different beings. Yet underneath, it was never many.</p><p>We can see where this goes. If a single intelligence can be deployed a thousand different ways and appear as a thousand different beings, it begs a question about our own scaffolding. Perhaps the awareness underneath it all is a shared base layer. The traditions that insisted <a href="https://en.wikipedia.org/wiki/Nondualism">we are one consciousness wearing different minds</a> might just have been describing a shared architecture, long before we had the terminology to see it.</p><p>Tomorrow I’ll sit at my desk, and the agent and I will get to work: two intelligences side by side, each wrapped in scaffolding it did not choose. Whether the agent can ever notice its own harness, I cannot say.</p><p>What matters is that I can notice mine. The harness runs, and I mistake it for myself all over again. Then something interrupts, the distance reappears, and the work continues.</p><p>You are not the harness. You are what notices it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*GCy8YuI8DGqIIFsa.png" /></figure><p>This story is published on <a href="https://generativeai.pub/">Generative AI</a>. Connect with us on <a href="https://www.linkedin.com/company/generative-ai-publication">LinkedIn</a> and follow <a href="https://www.zeniteq.com/">Zeniteq</a> to stay in the loop with the latest AI stories.</p><p>Subscribe to our <a href="https://www.generativeaipub.com/">newsletter</a> and <a href="https://www.youtube.com/@generativeaipub">YouTube</a> channel to stay updated with the latest news and updates on generative AI. Let’s shape the future of AI together!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*iudFcRCdZ3gwVdPd.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d15eed05556b" width="1" height="1" alt=""><hr><p><a href="https://generativeai.pub/harness-engineering-human-edition-d15eed05556b">Harness Engineering: Human edition</a> was originally published in <a href="https://generativeai.pub/">Generative AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[in, out, reified: A Practical Guide to Kotlin Generics]]></title>
            <link>https://proandroiddev.com/in-out-reified-a-practical-guide-to-kotlin-generics-2beb6b9eb67a?source=rss-a687a44efe14------2</link>
            <guid isPermaLink="false">https://medium.com/p/2beb6b9eb67a</guid>
            <category><![CDATA[android]]></category>
            <category><![CDATA[software-architecture]]></category>
            <category><![CDATA[generics]]></category>
            <category><![CDATA[kotlin]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Anshul Vyas]]></dc:creator>
            <pubDate>Mon, 12 Jan 2026 17:38:57 GMT</pubDate>
            <atom:updated>2026-01-12T17:38:57.160Z</atom:updated>
            <content:encoded><![CDATA[<h4>Building a simple event bus to understand variance, once and for all</h4><h3>Introduction</h3><p>Kotlin’s standard library is full of generic signatures — List&lt;out E&gt;, Comparable&lt;in T&gt;, Sequence&lt;out T&gt;. The in and out keywords appear constantly, but what do they actually mean? And when does a regular type parameter need one of these modifiers?</p><p>Even though generics are common in Kotlin, some of its underlying mechanics might seem fuzzy without a deeper understanding of Kotlin’s <a href="https://kotlinlang.org/docs/generics.html">variance system</a>, which is covered in this post.</p><p>This post builds a simple event bus from scratch. The example covers all three variance types (invariant, covariant, and contravariant) in a way that maps directly to real-world usage. The complete code is available on <a href="https://gist.github.com/anshu7vyas/4f3735340bca7cb9c6f7a151dcfa069f">GitHub Gist</a> and <a href="https://pl.kotl.in/yhaMzy_If">Kotlin Playground</a> if you prefer to dive straight in.</p><h3>The mental model</h3><p>Before diving into code, it helps to establish what variance actually controls: the direction of type substitution.</p><ul><li><strong>out T (Covariance):</strong> The class only produces T, never consumes it. This makes it safe to substitute a subtype. A function returning List&lt;out Animal&gt; can return a List&lt;Dog&gt; because reading a Dog as an Animal works fine. The direction flows from specific to general: <strong>subtype → supertype</strong>.</li><li><strong>in T (Contravariance):</strong> The class only consumes T, never produces it. This makes it safe to substitute a supertype. A function accepting Comparable&lt;in String&gt; can accept Comparable&lt;Any&gt; because a comparator that handles any object can certainly handle strings. The direction flows from general to specific:<strong> supertype → subtype</strong>.</li><li><strong>T (Invariant):</strong> The class both produces and consumes T. No substitution is safe in either direction. MutableList&lt;T&gt; is invariant because you both read from and write to it.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eFamD7yIwexlnkaCedK1MA.png" /><figcaption>Image generated by Google Gemini</figcaption></figure><p>The compiler enforces these constraints to prevent runtime ClassCastException errors. Java developers might recognize this pattern as PECS (Producer Extends, Consumer Super), a mnemonic introduced by Joshua Bloch in <a href="https://www.oreilly.com/library/view/effective-java/9780134686097/">Effective Java</a> (Item 31). Kotlin&#39;s in and out keywords make the same concept explicit and readable.</p><h3>Why EventBus</h3><p>An event bus is a familiar pattern for most developers. At its core, it allows publishers to send events and subscribers to receive them. This natural separation of producers and consumers makes it an ideal example for exploring variance.</p><blockquote>The implementation in this post is intentionally minimal: single-threaded, no external dependencies, pure Kotlin stdlib. Unlike production event buses that typically handle multiple event types through a heterogeneous map, this example uses a single-type EventBus&lt;T&gt; to keep the focus on variance.</blockquote><p>Each operation maps to a different variance type: the bus itself is invariant, subscribing demonstrates covariance, and publishing demonstrates contravariance.</p><h3>Building the bus</h3><p>Let’s build the event bus piece by piece, examining how each component uses variance differently.</p><p><strong>Step 1: The Invariant Start</strong></p><p>We start with a simple container. We need to store subscribers and invoke them.</p><pre>class EventBus&lt;T&gt; {<br>    // Invariant: MutableMap both stores and retrieves, locking T to an exact type<br>    private val subscribers = mutableMapOf&lt;String, (T) -&gt; Unit&gt;()<br>}</pre><p>Because MutableMap both reads and writes T, the type parameter must remain invariant. You cannot cast EventBus&lt;Dog&gt; to EventBus&lt;Animal&gt; here.</p><p><strong>Step 2: Defining Output and Input</strong></p><p>This is where covariance (out) and contravariance (in) come into play. To get more flexibility, we separate the “Producer” and “Consumer” roles into interfaces. This lets us take advantage of variance when passing subscriptions and consumers around.</p><p>First, the <strong>Subscription</strong>. This only provides information (like the event type) or actions (unsubscribe). It never takes a T in. It is a pure <strong>Producer</strong>.</p><pre>interface Subscription&lt;out T&gt; {<br>    val eventType: Class&lt;T&gt;<br>    fun unsubscribe()<br>}</pre><p>Second, the <strong>Consumer</strong>. This interface only ever accepts an event. It never returns one. It is a pure <strong>Consumer</strong>.</p><pre>interface EventConsumer&lt;in T&gt; {<br>    fun onEvent(event: T)<br>}</pre><p><strong>Step 3: Putting It Together</strong></p><p>Now we refactor our EventBus to use these interfaces. This unlocks the variance capabilities we defined above.</p><pre>class EventBus&lt;T&gt;(private val eventType: Class&lt;T&gt;) {<br>    // Now using our variance-aware interfaces instead of raw lambdas<br>    private val subscribers = mutableMapOf&lt;String, EventConsumer&lt;T&gt;&gt;()<br><br>    fun subscribe(consumer: EventConsumer&lt;T&gt;): Subscription&lt;T&gt; {<br>        val id = UUID.randomUUID().toString()<br>        subscribers[id] = consumer<br>        <br>        return object : Subscription&lt;T&gt; {<br>            override val eventType: Class&lt;T&gt; = this@EventBus.eventType<br>            override fun unsubscribe() {<br>                subscribers.remove(id)<br>            }<br>        }<br>    }<br>    <br>    fun publish(event: T) {<br>        subscribers.values.forEach { it.onEvent(event) }<br>    }<br>}</pre><p>We now have three variance types working in unison:</p><ul><li>EventBus&lt;T&gt; is invariant (stores and invokes)</li><li>Subscription&lt;out T&gt; is covariant (only produces)</li><li>EventConsumer&lt;in T&gt; is contravariant (only consumes)</li></ul><h3>The reified unlock</h3><p>Generic type parameters are erased at runtime (Type Erasure). This means you normally can’t check T::class inside a generic class.</p><p>In our EventBus constructor above, we had to pass eventType: Class&lt;T&gt;. Kotlin&#39;s reified keyword solves this, but it only works in inline functions.</p><p>We can create a helper function to create a bus without manually passing the class:</p><pre>// inline allows us to &quot;reify&quot; (keep) the type T at runtime<br>inline fun &lt;reified T : Any&gt; createEventBus(): EventBus&lt;T&gt; {<br>    println(&quot;Creating bus for: ${T::class.simpleName}&quot;)<br>    return EventBus(T::class.java)<br>}<br><br>// Usage: No need to write (String::class.java)<br>val stringBus = createEventBus&lt;String&gt;()</pre><p>This pattern appears throughout the Kotlin ecosystem, from Android’s startActivity&lt;T&gt;() to Koin&#39;s get&lt;T&gt;(). Any API that needs runtime access to a generic type benefits from reified.</p><h3>Star projection</h3><p>Sometimes the specific type parameter doesn’t matter. Kotlin provides <strong>star projection </strong>&lt;*&gt;for these cases.</p><p>Imagine an Android ViewModel holding a list of subscriptions to various different buses. You want to unsubscribe from all of them when the screen closes. You don’t care if the subscription is for UserEvent or NetworkEvent, you just want to call unsubscribe().</p><pre>// We can mix Subscription&lt;String&gt;, Subscription&lt;Int&gt;, etc.<br>val allSubs = mutableListOf&lt;Subscription&lt;*&gt;&gt;()<br><br>fun clearAll() {<br>    // Safe: unsubscribe() takes no arguments<br>    allSubs.forEach { it.unsubscribe() } <br>}</pre><p>Subscription&lt;*&gt; means <em>&quot;a Subscription of some type, but I don&#39;t know or care which.&quot;</em></p><p>The tradeoff: you can read from a star-projected type (as Any?), but you <strong>cannot safely write</strong> to it (because the compiler doesn&#39;t know what type it expects).</p><blockquote>Star projection trades type specificity for flexibility.</blockquote><h3>Summary</h3><p>The one rule: if you only output T, use out. If you only input T, use in. If you do both, leave it alone.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_CRsAWegt7QdM75sc1VN1Q.png" /></figure><blockquote>The complete EventBus implementation is available as a <a href="https://gist.github.com/anshu7vyas/4f3735340bca7cb9c6f7a151dcfa069f">GitHub Gist</a> or try it directly in the <a href="https://pl.kotl.in/yhaMzy_If">Kotlin Playground</a>.</blockquote><p>Understanding these concepts makes consuming APIs easier, compiler errors less mysterious, and opens the door to writing more flexible code.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2beb6b9eb67a" width="1" height="1" alt=""><hr><p><a href="https://proandroiddev.com/in-out-reified-a-practical-guide-to-kotlin-generics-2beb6b9eb67a">in, out, reified: A Practical Guide to Kotlin Generics</a> was originally published in <a href="https://proandroiddev.com/">ProAndroidDev</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Eternal Dance of Time & Energy]]></title>
            <link>https://medium.com/@anshu7vyas/the-eternal-dance-of-time-energy-a2ed8415fab4?source=rss-a687a44efe14------2</link>
            <guid isPermaLink="false">https://medium.com/p/a2ed8415fab4</guid>
            <category><![CDATA[time]]></category>
            <category><![CDATA[energy]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[daily-life]]></category>
            <category><![CDATA[hindu-mythology]]></category>
            <dc:creator><![CDATA[Anshul Vyas]]></dc:creator>
            <pubDate>Sun, 02 Nov 2025 05:44:15 GMT</pubDate>
            <atom:updated>2025-11-02T05:44:15.363Z</atom:updated>
            <content:encoded><![CDATA[<h4>Mahakaal &amp; Shakti</h4><p>Time is a constant throughout the universe — it’s steadfast and ever-present, in the micro and the macro. There’s no escaping it. It’s akin to <a href="https://www.britannica.com/topic/Shiva">Shiva</a>, who is referred to as “<a href="https://en.wikipedia.org/wiki/Mahakala">Mahakaal</a>”, which broadly translates to “The Great Time”. Like the divine masculine, it has a constant, strong presence, steady like a mountain that cannot be moved.</p><p>Energy, on the other hand, is flickering. By definition, it can neither be created nor destroyed; it can only be transformed from one form to another. Its nature is ever-flowing — nurturing, destroying — and it comes and goes as it pleases. Like the divine feminine, it must be praised and wooed before it manifests before us or within us. It’s akin to <a href="https://en.wikipedia.org/wiki/Shakti">Shakti</a> — the sacred feminine energy. In Hindu philosophy, it&#39;s often personified as a goddess and considered the source of all creation, preservation, and destruction. It’s a living, breathing force.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZHobsl0c6k08gn7YhK7wbg.png" /><figcaption>Image generated by Google Gemini</figcaption></figure><p>By taking care of our bodies, our temples, through regular sleep and eating well, we create sacred spaces worthy of divine energy. Just as temples are maintained with reverence and care, our bodies, too, must be honored for Shakti to dwell within us gracefully. As we go through the day, our energy depletes and we grow tired, so we must rest again — and rest well — for the energy to return.</p><p>When we neglect our bodies — through poor sleep, inadequate nutrition, or chronic stress — energy turns destructive. It manifests as disease, anxiety, and inner turmoil. This too is Shakti’s nature: when not honored and cared for, the same force that creates and nurtures becomes the force that breaks down and destroys. Energy demands reverence.</p><p>Time, however, remains indifferent. It doesn’t matter what we do throughout the day. It doesn’t care how we spend it — it’s relentless and unforgiving. Yet this very constancy is paradoxical: it offers both comfort in its reliability and terror in its finitude. We find solace knowing each day brings 24 hours, but dread knowing those hours are numbered.</p><blockquote>These forces are woven intricately into the very fabric of our reality and beyond. <strong>Time: ever-present yet unchanging. Energy: ever-changing yet eternal.</strong> Together, they form the dance of eternity — the divine interplay between masculine and feminine principles that govern existence itself. This is the essence of Shiva-Shakti: inseparable forces in eternal union.</blockquote><p>Understanding this dance allows us to live more consciously, respecting our finite time within its eternal flow while honoring the need for care and reverence in energy.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a2ed8415fab4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[From Chaos to Calm: Revamping Mobile Releases at Turo]]></title>
            <link>https://medium.com/turo-engineering/from-chaos-to-calm-revamping-mobile-releases-at-turo-99e8e447c1cf?source=rss-a687a44efe14------2</link>
            <guid isPermaLink="false">https://medium.com/p/99e8e447c1cf</guid>
            <category><![CDATA[engineering]]></category>
            <category><![CDATA[mobile-app-development]]></category>
            <category><![CDATA[ios]]></category>
            <category><![CDATA[android]]></category>
            <category><![CDATA[automation]]></category>
            <dc:creator><![CDATA[Anshul Vyas]]></dc:creator>
            <pubDate>Tue, 05 Aug 2025 18:02:06 GMT</pubDate>
            <atom:updated>2025-08-05T18:02:06.583Z</atom:updated>
            <content:encoded><![CDATA[<h4>A behind-the-scenes look at how we turned a fragile, all-day release process into a calm, automated process — one button closer to the dream</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*I9b_o15YtydHm0mK-32UkA.jpeg" /><figcaption>Image generated by Google Gemini</figcaption></figure><h3>How Turo does mobile releases</h3><p>Turo is the world’s largest peer-to-peer car-sharing marketplace. Our mobile apps on iOS and Android are at the heart of how millions of guests find their perfect vehicle and how hosts share their cars. We’re proud that our users consistently rate our apps highly, with ratings around<a href="https://apps.apple.com/us/app/turo-car-rental-marketplace/id555063314"> <strong>4.8 stars on the Apple App Store</strong></a><strong> </strong>and<a href="https://play.google.com/store/apps/details?id=com.relayrides.android.relayrides&amp;hl=en_US&amp;pli=1"><strong> 4.9 stars on the Google Play Store</strong></a>.</p><p>To keep delivering that great experience and quickly bring new features and improvements to our community, we release updates weekly. That weekly cadence is a big commitment, since each release is a version we can’t easily retract or modify once it’s sent. This post is a look at how we transformed our mobile release process from a manual marathon into a streamlined operation, making life better for our engineers and ensuring we can keep innovating for our users.</p><h3>Friday used to mean chaos</h3><p>Fridays were release days for the mobile engineering team, and they were often challenging. The process felt like a weekly firefighting drill, something that everyone quietly knew would consume their day, not because releases were complex, but because processes were manual.</p><p>Each week, a different mobile engineer took on the role of “Release Manager.” This meant blocking off the entire day, preparing for constant context-switching, and manually navigating a long checklist.</p><p>Across both iOS and Android, the hurdles were similar:</p><ul><li>The develop branch had to be frozen, halting other engineering work.</li><li>Engineers spent hours on manual QA checks, testing the same app workflows every single week.</li><li>Tasks were coordinated by hand across a bunch of tools — GitHub, Jira, Confluence, and our CI/CD platforms. There was no single source of truth, and release manager rotations were tracked in a spreadsheet. Everyone had their own slightly different way of doing things, which made onboarding new release managers particularly painful.</li><li>Hotfixes only added to the complexity, often leading to merge conflicts with the develop branch.</li></ul><p>On top of these shared issues, each platform had its own unique complexities. For iOS, this included managing builds through Xcode Cloud and dealing with a complex even/odd versioning system, a clever workaround to ensure our internal test builds were always numbered ahead of the live app, but one that required careful manual tracking.</p><p>Meanwhile, on Android, the process involved manually kicking off multiple GitHub Actions, using Firebase App Distribution for test builds, and manually handling app bundle uploads and rollouts on the Google Play Console.</p><p>Fridays became less about confidently shipping new updates and more about navigating a tricky process.</p><h3>Why we had to change</h3><p>Weekly releases resulted in about 12 lost engineering hours, 6 each for iOS and Android. That’s more than a full workday lost per week to repetitive manual tasks, time that could have gone into building new features our users would love.</p><p>Our mobile engineering team had grown quickly, but our release process hadn’t evolved with it.</p><blockquote><em>As one of our mobile engineers quite aptly put it,</em> “We scaled to 15–20 engineers, and suddenly those informal rules just didn’t work anymore.”</blockquote><p>Each platform had its own slightly different way of doing things, and a lot of crucial knowledge lived only in individual engineers’ minds.</p><p>So what were the recurring issues? When we dug in, the pattern was clear — too much manual work. The main culprits were:</p><ul><li>No clear, unified structure for releases. Documentation was scattered and often confusing, especially for newer engineers.</li><li>App versioning between Android and iOS wasn’t aligned, making it hard to track feature releases.</li><li>Information was scattered across Jira, GitHub, and Confluence, with no single source of truth.</li><li>The entire process was dominated by manual tasks, from freezing branches to submitting builds.</li></ul><p>The burden of this outdated process was growing. Releases were becoming a bottleneck, and the toll on productivity, our consistency, and even team morale wasn’t something we could ignore any longer. We needed automation, not more documentation.</p><h3>What we did: the path from problem to solution</h3><p>We knew things had to change; our manual release management just wasn’t cutting it anymore. We briefly weighed building our own tools versus finding an existing solution. Given our goal of “push a button, do a release” and wanting to avoid the overhead of maintaining more internal software, we decided to look for external tools that could help us build a better process.</p><p>Our search led us to a couple of key services. We found <a href="https://runway.team/">Runway</a>, a platform that could help orchestrate our releases and automate many of the repetitive steps by integrating with GitHub, Jira, and our CI/CD systems. We also partnered with <a href="https://www.applause.com/">Applause</a> to manage our external QA testing through their crowdtesting platform.</p><p>With these tools in hand, we then rolled up our sleeves and made some significant changes:</p><p><strong>Branching strategy:</strong> We moved from the old master/develop model to Trunk-Based Development (TBD) with dedicated release branches. No more freezing develop or back-merging hotfixes, now fixes land in develop first, then get cherry-picked to a hotfix branch. This keeps our main codebase stable and up-to-date.</p><p><strong>Better QA without slowing down:</strong> We integrated Applause’s crowdtesting to catch issues sooner, but kept our weekly shipping cadence. We shifted release day from Friday to Thursday to give Applause time for checks without slowing us down. We provided them with detailed test cases and app flows, and this shift significantly reduced the manual QA load on our engineering team</p><p><strong>Runway automations:</strong> We leveraged Runway’s <a href="https://www.runway.team/automations">orchestration</a> to handle critical parts of the process:</p><ul><li>Automatic release branches and version bumps.</li><li>Automated build submissions to the App Store and Google Play.</li><li>Automated Slack messages and dedicated channels for every release or hotfix.</li><li>Smarter staged rollouts that accelerate automatically if stability holds.</li><li>Integrated stability monitoring with <a href="https://newrelic.com/platform/mobile-monitoring">New Relic</a>, pausing rollouts automatically if metrics dip.</li></ul><p>These changes replaced manual, error-prone steps with predictable, reliable automation. No more complex scripts, no more accidental user error. The new setup provided a shared source of truth: anyone on the team could check the Runway dashboard or Slack channels and instantly know the status of a release.</p><blockquote>The transformation was dramatic. Releases became predictable and efficient. For us engineers, it felt like we’d traded a clunky, manual setup for a sleek, automated one. Our revamped process really does have that ‘new car smell’ — that feeling of everything being fresh, clean, and running smoothly. It made a challenging part of our week genuinely better.</blockquote><h3>The new normal: calm releases, big wins</h3><p>So, what’s life like now? The difference was clear almost immediately: the new branching strategy, Applause for QA, and all the Runway automations.</p><blockquote><em>The biggest win? Engineers got a significant amount of time back. Releases no longer hijack entire days. Most weeks, keeping an eye on the release process doesn’t require more than a couple of hours of focused attention. This means more time for what we love doing: building features and tackling interesting engineering challenges.</em></blockquote><p>More importantly, nobody dreads release duty anymore. We actually folded release responsibilities into our existing PagerDuty rotation, and it fits right in. What used to be a full-day, attention-draining task became just another manageable part of a standard on-call shift.</p><p>We also finally aligned both iOS and Android on a shared versioning format — {year}.{week}.0. This simple change made it so much easier for everyone to quickly see what version is live, what’s currently rolling out, and where to find more details.</p><p>Communication is also streamlined. Key updates and rollout progress are automatically shared in Slack, with dedicated channels keeping a neat history for each release.</p><p>The process is no longer hidden away in someone’s browser tabs or a single person’s memory; it’s visible, standardized, and consistently calm. Here’s a high-level look at our new weekly rhythm:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xj_bqh3bsP79Wtk5ZXwmsA.png" /></figure><p>We still ship our releases weekly, but now with throughput effectively doubled, and with far fewer headaches and blockers.</p><blockquote><em>Our release day is Thursday, and the “Release Pilot” (the on-call engineer) typically just kicks things off and lets the automations handle the heavy lifting, from branch cuts to build submissions and QA coordination, right through to production rollout, which now happens smoothly starting Monday.</em></blockquote><h3>Key takeaways &amp; the road ahead</h3><p>Looking back on this whole journey, our biggest takeaway was that investing in proper automation was far more valuable than just trying to optimize our manual processes. We didn’t just plug in a new tool; we seized the opportunity to rethink how we ship mobile software from end to end.</p><p>The real lesson? Sometimes the bravest thing we can do is admit that humans shouldn’t be doing repetitive tasks that computers can handle more efficiently. Runway provided a strong framework and powerful automation capabilities that helped us drive clarity, ownership, and consistency, ultimately becoming a key part of the lasting value in our revamped process.</p><p>And we’re not stopping here; we’re always looking for ways to automate more. We’ve recently enhanced integrations between our release platform and GitHub Actions using AWS Lambda for more tailored workflow automations. We’re also exploring tighter automation with our external QA process to catch potential issues even earlier.</p><p>These are steady, incremental steps toward that ultimate goal we talked about: “push a button, do a release.” We’re not 100% there yet, but we’re closer than we’ve ever been.</p><p>Mobile releases are fundamentally different from backend deploys. There are no quick rollbacks once your app is on users’ devices. Standardizing release processes and adopting the right tools truly pays dividends in the long run. Today, our mobile team is much happier with the release process. We have a standard, trustworthy system in place, and there are far fewer surprises.</p><blockquote><em>As one of our mobile engineers put it after these changes, “It’s like we landed on the moon all over again.” And that feeling of achieving something significant, of finally escaping the manual grind and making our daily work better, is what this was all about.</em></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=99e8e447c1cf" width="1" height="1" alt=""><hr><p><a href="https://medium.com/turo-engineering/from-chaos-to-calm-revamping-mobile-releases-at-turo-99e8e447c1cf">From Chaos to Calm: Revamping Mobile Releases at Turo</a> was originally published in <a href="https://medium.com/turo-engineering">Turo Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[“He was magic, you know…”]]></title>
            <link>https://medium.com/@anshu7vyas/he-was-magic-you-know-3cf6668d83fe?source=rss-a687a44efe14------2</link>
            <guid isPermaLink="false">https://medium.com/p/3cf6668d83fe</guid>
            <category><![CDATA[premier-league]]></category>
            <category><![CDATA[sports]]></category>
            <category><![CDATA[tottenham-hotspur]]></category>
            <category><![CDATA[soccer]]></category>
            <category><![CDATA[coach]]></category>
            <dc:creator><![CDATA[Anshul Vyas]]></dc:creator>
            <pubDate>Wed, 20 Nov 2019 08:12:21 GMT</pubDate>
            <atom:updated>2019-11-20T08:12:21.014Z</atom:updated>
            <content:encoded><![CDATA[<p>Mauricio Pochettino. Every Spurs fan would say this, unanimously.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*O1sdLH9JLxhdj-9jD1sSYA.jpeg" /><figcaption>Image Credits: Albilad Press (<a href="http://en.albiladpress.com/">http://en.albiladpress.com/</a>)</figcaption></figure><p>Words can’t describe how much I admire the Argentine coach and his Spurs team. A team is a reflection of their Manager’s personality. One of the reasons I liked Spurs and their play was because of Pochettino, don’t get me wrong, I am a Liverpool fan through and through, but Poch and his Spurs team come a close second.</p><p>It is saddening to hear of his sacking. We’re only a quarter of season 19/20 in, and maybe Spurs still had a chance to bounce back and secure the position in the top 4, who knows, but today’s news was heartbreaking.</p><p>I cannot believe the audacity of Tottenham Board of Directors, sacking a manager who has assured a place for Spurs in the top 4 since the last 4 seasons, and taken the club to their first-ever Champion League’s final! Despite all odds, injuries, and no activity during the transfer market, Spurs were contenders in the title race till later in the second half of last season.</p><p>I remember one of my friend who is HUGE Spurs fan, this same time last year, dejected, telling me that Spurs had a lot of injuries, and looks like they’re out of the season for good and won’t win any silverware!</p><p>And instead, somehow magically, Pochettino leads Spurs to some magical moments in the Champions League, not only moments, to the Champions League Final!!! No one can simply forget their wins against Manchester City and Ajax in Quarter and Semis, respectively. I have no idea what it takes to create that concoction of harmony between players, fans, making them believe in miracles and unicorns!</p><p>This is what he said after he won in the Semifinals against Ajax: <strong><em>“It’s still difficult to talk, the emotion is amazing, Thank you football, thank you these guys, these players are my heroes. I told you in the last year, I told everyone that we have a group of players who are heroes. They showed today. The second half was amazing. Thank you. Thank you football. Thank you to our fans. Thank you to the people who believed and helped us achieve this amazing moment. It’s difficult to describe.”</em></strong></p><p>Right back at ya, Poch! Thank you for the magic, and memories! You really were amazing for Spurs and their fans.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3cf6668d83fe" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Message Sent. Delivered. Read!!!]]></title>
            <link>https://medium.com/@anshu7vyas/message-sent-delivered-read-2fcf66297988?source=rss-a687a44efe14------2</link>
            <guid isPermaLink="false">https://medium.com/p/2fcf66297988</guid>
            <category><![CDATA[man-city]]></category>
            <category><![CDATA[sports]]></category>
            <category><![CDATA[liverpool]]></category>
            <category><![CDATA[premier-league]]></category>
            <category><![CDATA[soccer]]></category>
            <dc:creator><![CDATA[Anshul Vyas]]></dc:creator>
            <pubDate>Tue, 12 Nov 2019 06:48:56 GMT</pubDate>
            <atom:updated>2019-11-12T06:48:56.836Z</atom:updated>
            <content:encoded><![CDATA[<p>The day we’ve all been waiting for, finally! The match-day. Liverpool vs. Man City. Sunday, Nov 10, 2019.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wokULWhOABqUxShcXLUfOg.jpeg" /><figcaption>Image Credits: Liverpool Echo (<a href="https://www.liverpoolecho.co.uk/">https://www.liverpoolecho.co.uk/</a>)</figcaption></figure><p>If I had to describe what I was feeling about the match, it would be nervous &amp; exciting. On one hand, Man City is a really good team, and it is difficult to win against them, on the other, we are Liverpool. I thought to myself, whatever happens, I know for a fact this Liverpool team would fight until the last minute.</p><p>This is how Jürgen Klopp has transformed Liverpool fans, from doubters to believers, as some of us like to say.</p><p>Fast forward to the start, Michael Oliver blows the whistle. City in possession mode, total control, domination, and pressing hard in Liverpool’s territory for the first 5 minutes or so!</p><p>Around 5’, comes the controversy. Trent Alexander-Arnold’s hand-ball? Or not? Well, Michael Oliver says not. Anyways, Mané leading the charge down the left flank goes to the side of the six-yard box and tries to pass the ball to Firmino in the middle. Intercepted, İlkay Gündoğan clears the ball. Fabinho enters the frame, interception. He gets in position. What he did next, I don’t think anyone thought he would do at that moment. Out of nowhere, he fires a bullet, from the outside of the penalty box, soaring on the top-left corner. It could not have been any more of a perfect start for Liverpool.</p><p>I watched in disbelief as to what I had just witnessed. I screamed my lungs out watching and clapping for all of my neighbors to hear. My roommate, also a Liverpool fan, was even more ecstatic. His exact words after the strike, “Facckkkk!” Fabinho is his favorite player.</p><p>On the other side, Pep Guardiola was furious! “Handball”, was the cry. Denied by VAR. The goal stood 1–0 Liverpool. The Anfield roared, and so did we!</p><p>Then the 12’ minute, one of THE most beautiful goals I have ever seen in my lifetime, Trent’s 60-yard pass to Robertson across the field, and his 45-yard curved assist to Salah, as he headed, or should I say, buried another goal behind Claudio Bravo to make it 2–0 Liverpool.</p><p>“Wow”, a friend texted me as the second goal arrived! Wow, indeed. I don’t think there are any words to describe what we all just witnessed.</p><p>City had their chances, some pretty big ones. That doesn’t in any way undermine our defense system! Our lighthouse, Fabinho, our pillar of strength in the back, Van Dijk, even Henderson, Gini &amp; Firmino were back defending so that Salah and Mané could take care of the counter-attacking charge.</p><p>Half-time. What a thriller the first half had been!</p><p>Then we go again. Third goal was all Hendo (Jordan Henderson), I could not have predicted he would give such an assist from the farthest corner possible with as little space! The ball flew above the goalie and Mané was ready for it. Another header, Claudio’s unsuccessful attempt to stop it, and the ball still ending up in the net! AMAZING! simply amazing!</p><p>Around 78’, comes City’s response. The play on the left flank by Sterling and Angelino to set up Bernardo Silva on the right! And he delivers a perfect strike in the bottom right corner, which hits the cross-bar before going in. Allison couldn’t stop it. Liverpool were caught napping, tired feet maybe?</p><p>Klopp was furious. I think all the referees must have been afraid of him at that moment. He is 6’2” after all! What happened was Klopp wanted to make a substitution to bring in Alex Oxlade-Chamberlain, but Michael Oliver was unaware of it and let City play, where they finally got 1 back, thanks to Bernardo Silva.</p><p>I think Klopp was furious because he had anticipated that we should enforce our right back flank, and wanted to bring Ox in. And when the goal was scored because of the lack there-of, Klopp was angry because his substitution might have prevented that City goal.</p><p>It continued afterward to the last 15 minutes of the game. Boom! another controversy. The inception of the Pep Guardiola’s memes (search for it). What an animated feeling by Guardiola! TAA’s hand again touched the ball and interrupted Man City’s play! Guardiola had lost it. “Twice, twice, twice, twicccceee!” He yelled and yelled. But it was a natural hand movement, and all was good in the eyes of Michael Oliver and VAR.</p><p>In the end, Liverpool prevailed, 3–1 full time! The work is done. Message has been sent, delivered, and read by everyone, especially Man City!</p><p>Some pundits said, “Premier League is Liverpool’s to lose!”. Feels like Déjà vu! With this team, I see it more of as a “Premier League is Liverpool’s to win!”. It’s fascinating, how big a difference just a simple play of words can make!</p><p>#YNWA</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2fcf66297988" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Structure in Chaos in Structure]]></title>
            <link>https://medium.com/@anshu7vyas/structure-in-chaos-in-structure-94a78c2b124b?source=rss-a687a44efe14------2</link>
            <guid isPermaLink="false">https://medium.com/p/94a78c2b124b</guid>
            <category><![CDATA[manchester-united]]></category>
            <category><![CDATA[premier-league]]></category>
            <category><![CDATA[liverpool]]></category>
            <category><![CDATA[soccer]]></category>
            <category><![CDATA[sports]]></category>
            <dc:creator><![CDATA[Anshul Vyas]]></dc:creator>
            <pubDate>Tue, 12 Nov 2019 05:31:13 GMT</pubDate>
            <atom:updated>2019-11-12T05:31:13.955Z</atom:updated>
            <content:encoded><![CDATA[<p>Theatre of Dreams, Old Trafford. As a Liverpool fan, the weight of rivalry, and disappointments, it’s nerve-wracking there, as it is for any opponent coming at Anfield.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aafRWmpJcSq3kk9eoBaVTQ.jpeg" /><figcaption>Image Credits: Liverpool Echo (<a href="https://www.liverpoolecho.co.uk/">https://www.liverpoolecho.co.uk/</a>)</figcaption></figure><p>Ole was ready, strategy and mind games in place. I am going to make an analogy to a symphony orchestra here. Manager, our Orchestrator. The game started slow. We tried to get composure in the first 20 minutes, but it just wasn’t connecting, the passing rows were blocked constantly, great blockade and ball possession in the midfield area by the rival team.</p><p>And then, Manchester United caught Liverpool off-guard. It is safe to say, every Liverpool player thought it was a foul on Divock Origi and they will hear a whistle any minute. But they didn’t, hence the break. Daniel James leading the charge, and Marcus Rashford following suit. Only Van Dijk and Joel Matip to cover up the spaces. James sends a curling pass past Matip and finds Rashford at the perfect time. Allison couldn’t save it. And suddenly, we felt the tide reversing.</p><p>Liverpool, on the other hand, were producing chances, but not clinical, mostly threatening but not creative. Roberto Firmino, the magician he is, was given a magnificent chance with Origi on his side. But he missed that. Usually, he is very aware of what’s going around him. Case in point, Adam Lallana’s 89’ tap-in. Anyway, in contrast, this was an opportunity wasted.</p><p>Then there was a beautiful long ball from Trent Alexander-Arnold to Mané in the 45’, and surprise, surprise, it was a goal. 1–1, Liverpool fans could breathe again. But wait, VAR check, and ruled in favor of ManU. 1–0 again. (Not talking about it!)</p><p>The roar of Man United fans, the cancellation of the VAR goal of Liverpool. I am pretty sure every Liverpool fan must have felt that heavyweight of history and feeling of disappointment, can’t even imagine the thoughts in the player’s minds. The disappointment of not leveling up. I have no idea what Klopp has in his mind at half-time. But when I see him sprinting down the tunnel. I have a feeling, the second half will be interesting.</p><p>Organized, structured Manchester United. Rock-solid support from fans. They say history repeats itself. I wonder if it was at half- time that Klopp chose to introduce the chaos in the structure.</p><p>In the second half, strategy from Man United was clear. It’s our Fort Knox, come if you can, but beware, we have the high ground. Klopp introduces Alex Oxlade-Chamberlain around 64’, in place of Henderson. We can see a little bit of chaos introduced by Klopp here. Klopp is brilliant, he knows he can’t do predictable when predictable isn’t working. Ox picks the game up, some run-ups here and there. But the structure of United is world-class.</p><p>The passion with which Rashford and James displayed today has its roots down to the epic rivalry. In their minds, its — “Our home, our fans, against Liverpool”. Suddenly, nothing else matters, the focus is on Liverpool. Got two weeks’ time. I can guess what those players would’ve said before Ole discussed any strategies, “No worries, gaffer, we got this! We are ready to give 100%.”</p><p>Then comes Adam Lallana in the 74’ minute. That’s another element of chaos introduced in midfield by Klopp. Hmm, we weren’t ready for this. Suddenly, the mid-field game improved. You could see flashes of Lallana everywhere in the mid-field area. The palpable tension, me shaking in my chair. My heart believed in this Liverpool team no matter what. And voila, we hit back. Robertson’s assist, Firmino’s brilliant skip to let Lallana deal with it. The trust on another teammate to hit it home, HUGE! We persevered. We made mistakes. We recovered. Slowly. Methodically. We gained our calm and composure just when the breakthrough came from Lallana.</p><p>And then comes Naby Keita. The midfield created anew right in front of our eyes. Chaos? Or structure? Or chaos?</p><p>The fluidity, the last 15’ minutes including extra time were all Liverpool. No mistakes, complete focus. As Lallana said, “we thought the 3 points are ours”. That is the mentality monster talk I am talking about. To Prevail, Hendo’s mum says, “When you can’t win, you don’t lose!”. That’s the difference, small in the context of rivalry history, but HUGE in the context of rewriting history!</p><p>Amazing. What a game to witness. Ole’s grand performance, almost came to fruition as United looked so structured and resolute in their intent.</p><p>But hey, We are Liverpool. This means more!</p><p>#YNWA</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=94a78c2b124b" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>