<?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 Rituraj kumar on Medium]]></title>
        <description><![CDATA[Stories by Rituraj kumar on Medium]]></description>
        <link>https://medium.com/@krishnarituraj17?source=rss-6e7da8233bb9------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*p5_aP8oOEzCqrnXRNXmtZA.png</url>
            <title>Stories by Rituraj kumar on Medium</title>
            <link>https://medium.com/@krishnarituraj17?source=rss-6e7da8233bb9------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 23:36:03 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@krishnarituraj17/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[Augmented Reality Chatbots]]></title>
            <link>https://medium.com/zeals-tech-blog/augmented-reality-chatbots-4cbec463af6?source=rss-6e7da8233bb9------2</link>
            <guid isPermaLink="false">https://medium.com/p/4cbec463af6</guid>
            <category><![CDATA[webar]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[augmented-reality]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[webxr]]></category>
            <dc:creator><![CDATA[Rituraj kumar]]></dc:creator>
            <pubDate>Fri, 03 Jun 2022 08:05:19 GMT</pubDate>
            <atom:updated>2022-06-03T16:36:17.632Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>“Power of Augmented Reality technologies with Chatbots”</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ATlRZIx9HyHIYSXqBANDTw.jpeg" /></figure><h3>Introduction</h3><p>Hi there, In this post, I will be giving an overview about the Augmented Reality Chatbot use case and provide an end-to-end implementation solution for the same. Before we jump to the main part just take a look at the sneak peak real-time example of the Augmented Reality chatbot by <a href="https://www.nextechar.com/">NexTech AR Solutions.</a></p><h3>Sneak peek at Real Time Application example</h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F3nS0vCyT3eE%3Fstart%3D9%26feature%3Doembed%26start%3D9&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D3nS0vCyT3eE&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F3nS0vCyT3eE%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/2d02c137393bd49ea6000cf0dc62f297/href">https://medium.com/media/2d02c137393bd49ea6000cf0dc62f297/href</a></iframe><h3>What is Augmented Reality (AR)?</h3><p><em>Augmented reality is the real-time use of information in the form of text, graphics, audio, and other enhancements integrated with real-world objects.</em></p><h4>Why it’s important?</h4><p><em>Till 2020, </em><a href="https://arvrjourney.com/top-17-augmented-reality-statistics-facts-and-trends-for-2021-and-beyond-ee94e8bb54b6"><em>Research</em></a><em> has shown that 10X marketers that employ AR see a 10x increase in engagement time, increasing retention and effectiveness. As such, 7 out of 10 media planners want to use AR to boost advertising and customer engagement.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*POjrOQYKoBXWsENq1fJ0og.jpeg" /></figure><h3>What is Virtual Reality?</h3><p>Virtual Reality (VR) is <strong>a computer-generated environment with scenes and objects that appear to be real, making the user feel they are immersed in their surroundings</strong>. This environment is perceived through a device known as a Virtual Reality headset or helmet.</p><ul><li><strong>Comparison of Augmented Reality vs Virtual Reality</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GyzuMyBuMGIdEDL_DFOYKQ.png" /></figure><ul><li><strong>Application Example of Augmented Reality and Virtual Reality</strong></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EFsd31E9hFxfla9UzfQSzw.png" /></figure><h3>ADVANTAGES OF AUGMENTED REALITY</h3><p>AR Digital services and products are helping enterprises create touchless digital experiences, which allow the customer to understand precisely how and where a product fits in their business</p><ul><li>Creates unique customer experiences</li><li>Eliminates cognitive overload</li><li>Heightens user engagement</li><li>Competitive differentiation</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*b09UfYj1iGQHQrWsWbLV6Q.jpeg" /></figure><h3>Augmented Reality Usecase consumer Stats</h3><p>Snap Inc. reports that there are already more than 100 million consumers shopping with AR online and in stores.<a href="https://forbusiness.snapchat.com/blog/the-next-inflection-point-more-than-100-million-consumers-are-shopping-with-ar"> In 2021, the company teamed up with Deloitte Digital to survey 15,000 consumers worldwide</a>. While most respondents said they currently consider AR a “toy” (think: Snapchat filters, Pokémon Go, or Google’s AR-enabled search, which, at one point, had everyone placing dinosaurs in their home), a whopping 76% want to use it as a “tool” in their everyday lives.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/470/1*v0ChBoxZhcdH8uQyb3Y_vQ.png" /></figure><p>So, what are the most realistic and practical uses for it that already exist today? Online shopping is one of the most practical ones. But there are many more interesting sectors that are also bringing AR technology. Some of them I have explained in the <strong>Application Usecases of Augmented Reality with Chatbot </strong>section of this article.</p><p>According to Google’s 2019 AR Survey, <a href="https://www.thinkwithgoogle.com/consumer-insights/consumer-trends/ar-shopping-interest-statistics/">66% of people are interested in using AR</a> for help when making purchasing decisions. The pandemic has only accelerated this process. If those same surveys were conducted today, the numbers would likely be much higher.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RX5YzoHXN_W9wwcwrw0kXg.png" /></figure><h3>Application Usecases of Augmented Reality with Chatbot</h3><h4>AR in Advertising</h4><p>AR advertising is a mobile ad unit that uses the smartphone camera to superimpose 3D assets, such as game characters and scenarios from the advertised game, onto the user’s real-world surroundings.</p><p>AR ads can run on both iOS and Android in-app traffic, within rewarded video and display placements.</p><p><strong>Company examples</strong>:<a href="https://arinsider.co/2020/05/06/how-will-ar-transform-advertising/">[Link]</a></p><ul><li>Samsung</li><li>Coke</li><li>Nike</li><li>Others</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TuV-KlUlFVXZxl8pvRwvZA.jpeg" /></figure><h4>AR in Skincare and Haircare</h4><p>AR helps in recognizing how confusing it can be for customers to pick the right skincare products, SAL provides them a more in-depth insight into their skin type and recommends the most effective products.</p><p>Customers need to upload a selfie and complete a short quiz on SAL to receive a personalized skin diagnosis and product recommendations within a minute.</p><p><strong>Company examples </strong>[<a href="https://botcore.ai/blog/augmented-reality-chatbots/">Link</a>]</p><ul><li>Ponds</li><li>Sephora</li><li>Madison Reed</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*PKcKzlmUVNU6hi_bVb2C9A.jpeg" /></figure><h4>AR in Shopping</h4><p>AR helps the customers to get a dynamic and immersive experience while shopping for furniture, clothes, electronics products, and it also keeps them engaged and convinces them about the utility of the purchase.</p><p><strong>Company examples </strong>[<a href="https://www.screenmedia.co.uk/lab/the-future-of-online-shopping-how-chatbots-ar-and-voice-are-disrupting-ecommerce/">Link</a>]:</p><ul><li>Shopify</li><li>Nike</li><li>Ikea</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6H06R-tL9hfi1SJudXpBmQ.png" /></figure><h4>AR in Enterprise Solution</h4><p>Enterprise augmented reality project is likely of strategic importance to each organization. Every organization has different usecases in mind for the integration of AR and VR technology into their ecosystem of products and services. Some of the open-source example use cases are given below</p><ol><li>Measure spaces and objects around you in augmented reality. <a href="https://experiments.withgoogle.com/measureup">Built by Google Creative Lab using the WebXR API.</a></li></ol><ul><li>Can be used for Validating whether your future car will fit in the parking space allotted to it or not.</li><li>Measure the product space availability.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*XzC7-1kqjZeotU1tvDNtFQ.gif" /></figure><p>2. Operator-machine communication use cases. <a href="https://www.onirix.com/top-5-uses-of-web-ar-and-its-impact-on-the-market/">Onirix Company has a lot of ready-to-market implementation of AR technology for many enterprise organizations for different use cases like tourism, media, Automobile, construction and many others.</a></p><ul><li>Operations, maintenance, review or repair, or security tasks.</li><li>Digitization through technologies such as AR allows to reduce times in the day to day, and therefore makes the operator more efficient in his work</li><li>Continuous help can be offered to the operator so that he can interact with his immediate environment, in which case it is usually quite complex, with a multitude of options and possibilities</li></ul><h3>Technology’s available to implement for AR and VR</h3><ul><li>JavaScript library</li><li><a href="https://ar-js-org.github.io/AR.js-Docs/">AR.js</a></li><li><a href="https://www.x3dom.org/">X3DOM</a>.</li><li><a href="https://modelviewer.dev/">ModelViewer</a></li><li><a href="https://developers.google.com/ar">ARCore</a></li><li><a href="https://developer.apple.com/augmented-reality/">ARKit</a> (Specific to iOS devices)</li><li><a href="https://unity.com/unity/features/ar">Unity core ( Unified workflow platform)</a></li><li><a href="https://www.ptc.com/en/products/vuforia">Vuforia</a></li><li><a href="https://sparkar.facebook.com/ar-studio/">Spark AR (for FB and Instagram).</a></li></ul><h3>Demo Time</h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2Fd9O5lPyufbSfZzC1pq%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2Fd9O5lPyufbSfZzC1pq%2Fgiphy.gif&amp;image=https%3A%2F%2Fi.giphy.com%2Fmedia%2Fd9O5lPyufbSfZzC1pq%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="244" frameborder="0" scrolling="no"><a href="https://medium.com/media/3ce1c9342149402640552aa7de580e7b/href">https://medium.com/media/3ce1c9342149402640552aa7de580e7b/href</a></iframe><p>In this demo, I will demonstrate how to use the AR Chatbot usecase on Instagram.</p><p>It has been implemented on Instagram with the help of the <a href="https://manychat.com/">ManyChat</a> tool and open-source WebXR examples</p><blockquote><strong>Note:</strong> I have not developed any WebAR page in this demo. I have used open sources projects which are already available.</blockquote><p><strong>Open source WebXR examples:</strong></p><ol><li>Measure spaces and objects around you in augmented reality b<a href="https://experiments.withgoogle.com/measureup">uilt by Google Creative Lab using the WebXR API.</a></li><li>AR Shopping example created by the <a href="https://www.marvinai.live/playground">Marvin XR team</a></li></ol><blockquote>Check the Complete Flow Diagram<strong><em> </em></strong><a href="https://manychat.com/flowPlayerPage?share_hash=426575_00688d24443bd21c95d80386a8d812dbff764d20"><strong>Here</strong></a> for the <strong>Augmented reality chatbot FLow Builder using Manychat.</strong></blockquote><h4>Pre-requisite to Interact with the demo Instagram Chatbot</h4><ul><li>A smartphone that supports VR and AR</li><li>Instagram account</li><li>Google chrome browser for most of the WebXR features to function properly.</li></ul><h4>Steps to access the demo Instagram chatbot:</h4><ol><li>Open the Instagram App and login into your account</li></ol><p>2. Search for <strong>test_chatbotwebar</strong> user/account or click on the below link to go directly to the demo Instagram account</p><ul><li><a href="https://www.instagram.com/test_chatbotwebar/">test_chatbotwebar</a></li></ul><p>3. Click on the message option after entering the <a href="https://www.instagram.com/test_chatbotwebar/">test_chatbotwebar</a> Instagram page</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3122Io9pTFpIPcfr3eo0mA.jpeg" /><figcaption>Source: Author. Print from my Instagram page</figcaption></figure><p>3. Click on the provide option <strong><em>Hi, I want to explore ARChat demo 🚀 option</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DMXRJPaeM0GH1loZXHf-zA.jpeg" /><figcaption>Source: Author. Print from my Instagram page</figcaption></figure><p>4. Click on the Yes ✅ button/ option to start the AR chatbot experience</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IHdu1OxkmTaTZFMAmkczyA.jpeg" /><figcaption>Source: Author. Print from my Instagram page</figcaption></figure><h3>Working Video of the demo</h3><ul><li><strong>AR ChatBot Demo</strong></li></ul><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F5nXA1F5xTGE%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D5nXA1F5xTGE&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F5nXA1F5xTGE%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/d50afea73b18e8b1717bb8249ced627d/href">https://medium.com/media/d50afea73b18e8b1717bb8249ced627d/href</a></iframe><h4>Usecase Demo</h4><ul><li><strong>Shopping Augmented reality</strong></li></ul><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FyOd_hTICSgk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DyOd_hTICSgk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FyOd_hTICSgk%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/befa6c121eeffbbb81b4e6bb2c73f72f/href">https://medium.com/media/befa6c121eeffbbb81b4e6bb2c73f72f/href</a></iframe><ul><li><strong>Measure spaces and objects around you in augmented reality.</strong></li></ul><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2Fn1luySHJ3ss%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3Dn1luySHJ3ss&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fn1luySHJ3ss%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/1b95cc3d983a2c8824a3ee9ca57c2c64/href">https://medium.com/media/1b95cc3d983a2c8824a3ee9ca57c2c64/href</a></iframe><h3>REFERENCEs &amp; DOCUMENT Links</h3><ul><li><a href="https://www.onirix.com/success-stories/">Success Stories of Augmented Reality experiences with Onirix</a></li><li><a href="https://manychat.com/">ManyChat software tool for creating the Instagram page</a></li><li><a href="https://manychat.com/flowPlayerPage?share_hash=426575_00688d24443bd21c95d80386a8d812dbff764d20">Flow Setup Design for Chatbot Builder</a></li><li>AR Shopping Examples created by the <a href="https://www.marvinai.live/playground">Marvin XR team</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4cbec463af6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zeals-tech-blog/augmented-reality-chatbots-4cbec463af6">Augmented Reality Chatbots</a> was originally published in <a href="https://medium.com/zeals-tech-blog">The Zeals Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[GCP Vertex AI Pipeline in Action]]></title>
            <link>https://medium.com/zeals-tech-blog/gcp-vertex-ai-pipeline-in-action-2108d04b0a65?source=rss-6e7da8233bb9------2</link>
            <guid isPermaLink="false">https://medium.com/p/2108d04b0a65</guid>
            <dc:creator><![CDATA[Rituraj kumar]]></dc:creator>
            <pubDate>Tue, 17 May 2022 08:40:17 GMT</pubDate>
            <atom:updated>2022-05-17T08:40:17.740Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/626/1*cZHeTPGzxe_Co8NiFBn6PA.jpeg" /></figure><h3>Introduction</h3><p>Hello there, this is Rituraj and I work as the Data and ML Engineer at <a href="https://zeals.co.jp/en/">Zeals</a>.</p><p>In my <a href="https://medium.com/zeals-tech-blog/adopting-mlops-methodology-in-zeals-a88612762895"><strong><em>previous post</em></strong></a>, I have discussed why <a href="https://zeals.co.jp/en/"><strong>Zeals</strong></a> is adopting MLOps Methodology in the organization and how we have selected GCP Vertex AI Platform as our one-stop solution for the MLOps operations.</p><p>In this post, I will be explaining the basis of the ML Pipelines, Kubeflow Pipelines, how to create them to fill your custom Machine Learning needs, and how to run them in Vertex AI Pipelines and analyze their results, with a small example code.</p><p>Let us begin!</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2F5zf2M4HgjjWszLd4a5%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2F5zf2M4HgjjWszLd4a5%2Fgiphy.gif&amp;image=https%3A%2F%2Fmedia0.giphy.com%2Fmedia%2Fv1.Y2lkPTc5MGI3NjExODhsMmNqYnZpYnNocjd3MTFkd2k2Mm96dzEwbTg2OGdwcGplMzQ1bSZlcD12MV9naWZzX2dpZklkJmN0PWc%2F5zf2M4HgjjWszLd4a5%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="181" frameborder="0" scrolling="no"><a href="https://medium.com/media/ceb4fd2380af661521a6e88c61bea3a3/href">https://medium.com/media/ceb4fd2380af661521a6e88c61bea3a3/href</a></iframe><h3>Why are ML pipelines useful?</h3><p>Before we dive in, let’s first understand why you would want to use a pipeline. Imagine you’re building out an ML workflow that includes processing data, training a model, hyperparameter tuning, evaluation, and model deployment.</p><p>Each of these steps may have different dependencies, which may become unwieldy if you treat the entire workflow as a monolith. As you begin to scale your ML process, you might want to share your ML workflow with others on your team so they can run it and contribute code. Without a reliable, reproducible process, this can become difficult.</p><p>With pipelines, each step in your ML process is its own container. This lets you develop steps independently and track the input and output from each step in a reproducible way. You can also schedule or trigger runs of your pipeline based on other events in your Cloud environment, like kicking off a pipeline run when new training data is available.</p><p><strong>Here is where the </strong><a href="https://cloud.google.com/vertex-ai"><strong>GCP Vertex AI </strong></a><strong>comes into the picture as a serverless and managed MLOps platform.</strong></p><h3>Google Vertex AI and Kubeflow:</h3><p>Kubeflow is a machine learning toolkit that provides a pipeline solution called Kubeflow Pipelines, built atop Kubernetes. Google introduced Vertex AI Pipelines because maintaining Kubernetes can be challenging and time-intensive. It’s a serverless product to run pipelines, so your machine learning team can focus on what they’re there to do instead of worrying about the infrastructure setup for their use cases.</p><p>Google Vertex AI supports two different types of pipelines:</p><ul><li>Kubeflow Pipelines using the Kubeflow SDK ≥ v1.8.9</li><li>TensorFlow Extended Pipelines using the TFX SDK ≥ v0.30.0</li></ul><h3><strong>Vertex AI Pipeline</strong></h3><p>Vertex AI Pipelines helps you to automate, monitor, and govern your ML systems by orchestrating your ML workflow in a serverless manner and storing your workflow’s artifacts using Vertex ML Metadata. By storing the artifacts of your ML workflow in Vertex ML Metadata, you can analyze the lineage of your workflow’s artifacts — for example, an ML model’s lineage may include the training data, hyperparameters, and code that were used to create the model.</p><p>A Vertex AI Pipeline consists of multiple components, where each component has one specific responsibility. The component contains the code that this step in the pipeline should perform.</p><p>There are two different component types:</p><ul><li><strong>Python Function-Based Components</strong><br>The most simple one. Just write a Python function.</li><li><strong>Container-Based Component</strong>s<br>Run anything that you can pack into a Container as a component. You can now implement your component in any language.</li></ul><p>A pipeline consists of multiple steps, and each step is defined by one component. With the <strong>@component</strong> decorator provided by the Kubeflow Pipelines SDK, we can create a component based on a Python function</p><p><strong><em>For example:</em></strong></p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/3c271c63b29e0fe9aafa0cb8ce8aaf5b/href">https://medium.com/media/3c271c63b29e0fe9aafa0cb8ce8aaf5b/href</a></iframe><p>The decorator compiles the function into a component specification YAML file. You can make it visible by adding an additional parameter to the component decorator.</p><h3>Artifacts and Parameters</h3><p>Artifacts and Parameters represent the data that components either use (input) or produce (output).</p><p>Passing data between components A and B is not everything. Artifacts are also necessary for the reproducibility of our pipeline. We need to understand the differences between pipeline runs to improve our models or to find issues. We can’t compare models if we don’t know what kind of data or hyperparameter got used. It’s important to track everything that goes in or out of our components.</p><p><strong>Parameters</strong> are used for small types of data you want to pass from component A to B likeint, float, str, bool, dict or list.</p><p><strong>Artifacts</strong> are used for larger or complex data and are stored in Google Cloud Storage, based on the pipeline root we defined during pipeline creation.</p><p>There are <a href="https://github.com/kubeflow/pipelines/blob/27b3511ad5e8283096555efb4f00d5ba5427af11/sdk/python/kfp/v2/dsl/__init__.py#L19">different types of artifacts</a>:</p><ol><li>Dataset</li><li>Model</li><li>Metrics</li><li>Visualization</li></ol><p>We’ll cover Metrics and Visualizations in detail later in this article. The focus now is solely on how to work with your data and models.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/0*041qR9GADisVWJYc.png" /></figure><p>Artifacts are shown in the Vertex AI Pipeline UI</p><p>For artifacts that do not fit into concepts like datasets or models you can use the generic <a href="https://kubeflow-pipelines.readthedocs.io/en/latest/source/kfp.components.html#kfp.components.OutputPath">outputPath</a>, and <a href="https://kubeflow-pipelines.readthedocs.io/en/latest/source/kfp.components.html#kfp.components.InputPath">inputPath</a>.</p><p>It is important to understand that artifacts are nothing more than a <strong>simple reference</strong>, a file path.</p><p>We will now use our above knowledge to create an example Kubeflow Pipeline and run it in Vertex AI Pipelines.</p><h3>UseCase</h3><p>We will be taking a very common and well-known use case which is to <strong>“<em>Predict the wine quality</em>”. </strong>The data is downloaded from <a href="https://archive.ics.uci.edu/ml/datasets/wine+quality">UCI Machine Learning Repository</a> @source [Cortez et al., 2009].<br>We are going to use a common supervised machine-learning algorithm such as Random Forest.</p><p>For a detailed description of the use cases please do take a look at the<strong><em> </em></strong><a href="https://www.kaggle.com/datasets/rajyellow46/wine-quality"><strong><em>Wine Quality Kaggle page.</em></strong></a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*kcqyWeglUqHKwKrB" /><figcaption>Photo by <a href="https://unsplash.com/@scottiewarman?utm_source=medium&amp;utm_medium=referral">Scott Warman</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>All the codes below for this article are ready to be used in a <a href="https://github.com/rituraj17/vertex_pipeline_wine_quality_usecase/blob/main/vertex_pipeline_wine_quality_usecase.ipynb"><strong><em>Jupyter notebook</em></strong></a>.</p><h3>Pre-requisite Environment setup</h3><ul><li>Enable the APIs if they are not enabled.</li></ul><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/0a19ef58de8a3c3618a62547c0d368cb/href">https://medium.com/media/0a19ef58de8a3c3618a62547c0d368cb/href</a></iframe><ul><li>Install the Kubeflow and GCP AI platform libraries and components if not installed.</li></ul><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/3279e8deb1255fbd08b890a91a4ae6db/href">https://medium.com/media/3279e8deb1255fbd08b890a91a4ae6db/href</a></iframe><ul><li>Setup global variables</li></ul><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/f9e6ff4895b88f93faf920221c251a08/href">https://medium.com/media/f9e6ff4895b88f93faf920221c251a08/href</a></iframe><ul><li>Import-Packages</li></ul><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/96d9c656c57b8cfc7918a8bfb54d3db1/href">https://medium.com/media/96d9c656c57b8cfc7918a8bfb54d3db1/href</a></iframe><h3>Pipeline steps :</h3><ul><li>Load the data</li><li>Training of model</li><li>Evaluation of model</li><li>Deploy</li></ul><h4>Load the wine data</h4><p>The first component is to get or process the data<strong>. </strong>The component uses a pre-build Python3.9 base image in which additionally I’ve added: pandas, sklean, and pyarrow.</p><p>The logic is simple, we read the<a href="http://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-white.csv"> <em>wine quality dataset</em></a><em>, w</em>e choose the <strong><em>best_quality </em></strong><em>as a </em>label, and we split the data into train and test datasets.<br>The configuration of the component is stored in a YAML file called <em>get_wine_data.yaml</em> that you can further use to run the component without rewriting the code. The YAML file is very useful for generating component templates.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/41a1e9a39b5591104eba1a42987cfcd6/href">https://medium.com/media/41a1e9a39b5591104eba1a42987cfcd6/href</a></iframe><h3>Train the model</h3><p>This component uses a pre-build Python 3.9 base image and takes into input the training dataset stored on Google Storage which has been generated from the previous step.</p><p>The model is stored in pickle format on Google Storage under the PIPELINE_ROOT path. Please keep in mind that the supported model format is: scikit-learn: model.joblib or model.pkl, XGBoost: model.bst,<br>Tensorflow model.pb.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/2813d0033277c429a3a61b08b66d3d1d/href">https://medium.com/media/2813d0033277c429a3a61b08b66d3d1d/href</a></iframe><h3>Evaluate the model</h3><p>The evaluation component relies on a pre-compiled Python 3 base image, in which we install using “pip” additional packages such as pandas and sklearn.</p><p>It takes in input from the test dataset, the trained model, and the predefined accuracy threshold. Then it computes the accuracy, the ROC curve, and the confusion matrix and checks if the accuracy is above the predefined threshold.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/2cb42def410a54eae2c59994878a187e/href">https://medium.com/media/2cb42def410a54eae2c59994878a187e/href</a></iframe><h3>Deploy the model</h3><p>This component uses the same base image as the previous components. It takes into input the project name, the region, the trained model, and the URL of the serving container.</p><p>To deploy a model we first create an endpoint, upload the model to Vertex AI models, and then we deploy the model to the endpoint.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/a3780506a2e27723071fb4ea62ccec06/href">https://medium.com/media/a3780506a2e27723071fb4ea62ccec06/href</a></iframe><h3>Create the pipeline</h3><p>The pipeline itself helps you to orchestrate the ml workflow in a serverless manner. It takes as input a few parameters like the URL containing the raw data, the API endpoint, and the project.</p><p>It uses the following operators as steps: data operation, train model operation, model evaluation operation, and deploy model operation.</p><p>The steps are executed sequentially. The deploy model operation step is executed only if the accuracy of the trained model is above the threshold. To do so, you need to use the conditional pipeline feature (<em>with dsl.Condition</em>).</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/c350be64d555d6dd65740100930a780c/href">https://medium.com/media/c350be64d555d6dd65740100930a780c/href</a></iframe><h3>Run the pipeline</h3><p>The pipeline compilation generates the <strong><em>pipeline-wine-quality-job.json</em></strong><br>Use the compiler and specify the<em> pipeline_func</em> and the <em>package_path</em>.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/e270d15400d404b3089b4514a5e70545/href">https://medium.com/media/e270d15400d404b3089b4514a5e70545/href</a></iframe><p>Create a run using the job spec file previously generated and Execute and Submit the pipeline job</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/52668c6711349d6fb938a2173b405430/href">https://medium.com/media/52668c6711349d6fb938a2173b405430/href</a></iframe><p>After submitting the pipeline an HTTPS link will be generated. Click on the link to see the status of the generated pipeline job</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jVDWG3WCdQCgmFraZ15cMQ.png" /><figcaption>Source: Author. Print from my GCP Console.</figcaption></figure><p>You can also check your metric plots. It would look like something like below</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*75-r6ZV-XC0hSsuj-S9JBw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*zwwLx79jjjt5FnL0Jz18RA.png" /><figcaption>Source: Author. Print from my GCP Console.</figcaption></figure><p>After the pipeline has run successfully the model will be uploaded to the <a href="https://cloud.google.com/vertex-ai/docs/model-registry/introduction"><strong>Vertex AI Model Registry</strong></a> and then its get deployed to the <a href="https://cloud.google.com/vertex-ai/docs/predictions/deploy-model-api"><strong>Vertex AI Endpoints.</strong></a></p><h4>Model stored in Vertex AI</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FD29dd_F9EJpU6iV7q7HCA.png" /><figcaption>Source: Author. Print from my GCP Console.</figcaption></figure><h4>Endpoint Deployed in Vertex AI</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*b8lyKYSdKJ42Cs8lYhfMHw.png" /><figcaption>Source: Author. Print from my GCP Console.</figcaption></figure><h3>Simulate real-time prediction</h3><ul><li>Get the endpoint id of the model deployed from the Vertex AI Endpoint Services through the GCP console UI or execute the below gcloud command.</li></ul><pre># Syntax <br>gcloud ai endpoints list --region=[&quot;region&quot;]  \<br>--filter=[&quot;endpoint name&quot;]</pre><pre># For this blog sample example usecase we will be using the below cmd<br>gcloud ai endpoints list --region=&quot;asia-northeast1&quot; \<br>--filter=&quot;wine_quality_endpoint&quot;</pre><p>The output of the above command will look something like the below image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CM40KAzm5QsBbD2S1r7gIQ.png" /><figcaption>Source: Author. Print from my Vscode Editor</figcaption></figure><ul><li>Use the Endpoint ID obtained from the previous step and pass it to the below code along with the GCP project id and the Endpoint Region</li></ul><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/9f587f4e46f01ae8ad8536545ae4ac0b/href">https://medium.com/media/9f587f4e46f01ae8ad8536545ae4ac0b/href</a></iframe><ul><li>After executing the above code you get the below output as prediction results from the endpoint</li></ul><pre># Wine Quality prediction [0 - bad quality ,1 - Good quality]</pre><pre>Prediction(predictions=[0.0, 0.0, 0.0, 1.0], deployed_model_id=&#39;5844089019582906368&#39;, explanations=None)</pre><h3>Bonus Features of GCP Vertex Pipeline</h3><p>The GCP Vertex Pipeline also has a feature to compare two different pipelines which are very important when you want to have a complete understanding of the different pipeline jobs.</p><p>We can compare the pipeline by going through the GCP Vertex Pipeline Console and selecting the pipeline job to compare and clicking on the compare option on the page.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/991/1*yJvHbyWGxm3svjU9uSFFXg.png" /></figure><h3>Conclusion</h3><p>Phew! That was quite a lot of Theoretical and Practical Knowledge Adventure. Give yourself a pat on your back. 👋</p><p>With Vertex AI, you will have an easier and better life with all things related to MLOps. There is still much more that can be done in Vertex Pipelines, we have just scratched the surface of it. We will be exploring more features of GCP Vertex AI services and will be implementing them in our use cases. We will be sharing out the findings in our future blogs.</p><p>Stay tuned for that next piece of blog! Thank you for reading, and stay safe!</p><h3>Reference:</h3><ul><li><a href="https://cloud.google.com/vertex-ai/docs/pipelines/introduction">Introduction to Vertex AI Pipelines</a></li><li><a href="https://www.kubeflow.org/docs/components/pipelines/sdk-v2/python-function-components/#understanding-how-data-is-passed-between-components">Kubeflow Pipelines Docs</a></li><li><a href="https://www.kaggle.com/datasets/rajyellow46/wine-quality">Wine Quality Kaggle Datasets and Notebooks.</a></li><li><a href="https://github.com/googleapis/python-aiplatform">Vertex AI SDK</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2108d04b0a65" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zeals-tech-blog/gcp-vertex-ai-pipeline-in-action-2108d04b0a65">GCP Vertex AI Pipeline in Action</a> was originally published in <a href="https://medium.com/zeals-tech-blog">The Zeals Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Adopting MLOps Methodology in Zeals]]></title>
            <link>https://medium.com/zeals-tech-blog/adopting-mlops-methodology-in-zeals-a88612762895?source=rss-6e7da8233bb9------2</link>
            <guid isPermaLink="false">https://medium.com/p/a88612762895</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[gcp]]></category>
            <category><![CDATA[mlops]]></category>
            <category><![CDATA[vertex-ai]]></category>
            <category><![CDATA[mlops-platform]]></category>
            <dc:creator><![CDATA[Rituraj kumar]]></dc:creator>
            <pubDate>Wed, 22 Dec 2021 10:51:55 GMT</pubDate>
            <atom:updated>2021-12-22T10:51:55.962Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*neHuSRLTkVyW9msj7E3NCQ.jpeg" /></figure><h3>Introduction</h3><p>Hello There, this is Rituraj and I work as the Data and ML Engineer at <a href="https://zeals.co.jp/en/">Zeals</a>.</p><p>If you want to learn about how Zeals is Adopting MLOps Methodology you have come to the right place.</p><p>Let us begin!</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2F5VhzCYV4lKVxaZbzmD%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2F5VhzCYV4lKVxaZbzmD%2Fgiphy.gif&amp;image=https%3A%2F%2Fi.giphy.com%2Fmedia%2F5VhzCYV4lKVxaZbzmD%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="243" frameborder="0" scrolling="no"><a href="https://medium.com/media/991bf80b93a4ee36386f7972debf7bfa/href">https://medium.com/media/991bf80b93a4ee36386f7972debf7bfa/href</a></iframe><h3>Background Story</h3><p>We have data scientists and operations professionals on our team, and while they do their best to collaborate, they each have their own areas of expertise.</p><p>This could lead to miscommunications and misunderstandings. The data scientists can interpret the data, but they likely don’t have the background to manage business operations. Likewise, the business team can make their side of things work, but they don’t fully know how to interpret and implement data.</p><p>Here is where machine learning operations (MLOps) and GCP Vertex AI comes in.</p><h3>What is MLOps?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HhdPlO5EkyndEvZbm8_wjA.jpeg" /></figure><p>MLOps (machine learning operations) is a set of best practices for improving communication and collaboration between the data science and operations team that aims to make developing and maintaining production machine learning seamless and efficient.</p><p>It is an extension of what we know as DevOps. DevOps is the process of organizing cooperation between everyone involved in the design and building of big data. This process has been around for a while, but MLOps is still in its youth.</p><p>The data science community generally agrees that it’s an umbrella term for best practices and guiding principles around machine learning — not a single technical solution.</p><p>The Four guiding Pillars of MLops</p><ul><li>Machine learning should be collaborative.</li><li>Machine learning should be reproducible.</li><li>Machine learning should be continuous.</li><li>Machine learning should be tested &amp; monitored.</li></ul><h3>Why MLOps Matters?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9p7VsoynR03dkFb-ODMGew.jpeg" /></figure><p>The goal of MLOps is to reduce technical friction to get the model from an idea into production in the shortest possible time to market with as little risk as possible.</p><p>Traditionally, machine learning has been approached from a perspective of individual scientific experiments predominantly carried out in isolation by data scientists. However, as machine learning models become part of real-world solutions and critical to business, we will have to shift our perspective, not to depreciate scientific principles but to make them more easily accessible, reproducible, collaborative, and most importantly to increase the speed at which machine learning capabilities can be released.</p><p>MLOps <a href="https://cloud.google.com/solutions/machine-learning/mlops-continuous-delivery-and-automation-pipelines-in-machine-learning">creates a lifecycle</a> and a set of practices that apply to the development of machine learning systems. This includes research, development, operations, and implementation</p><p>The process of brainstorming, developing, and implementing machine learning is extensive. Having a set of duplicatable processes to guide each project helps in many ways.</p><h3>Why did we need to adopt MLOps?</h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2F3cXmze4Y8igXdnkc3U%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2F3cXmze4Y8igXdnkc3U%2Fgiphy.gif&amp;image=https%3A%2F%2Fmedia3.giphy.com%2Fmedia%2Fv1.Y2lkPTc5MGI3NjExZDhheTM2cXc0dTcxdXd5eHd4Y3R0Nmd5YXhvZm45ejNkNGxyMWk0biZlcD12MV9naWZzX2dpZklkJmN0PWc%2F3cXmze4Y8igXdnkc3U%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="326" frameborder="0" scrolling="no"><a href="https://medium.com/media/dc5091c083842830c92387c92334b771/href">https://medium.com/media/dc5091c083842830c92387c92334b771/href</a></iframe><p>Creating an ML model that can predict what you want it to predict from the data you have fed is easy. <em>However, creating an ML model that is reliable, fast, accurate, and can be used by a large number of users is difficult.</em></p><p>The necessity of MLOps can be summarized as follows:</p><ul><li>ML models <strong>rely on a huge amount of data</strong>, difficult for a single person to keep track of.</li><li><strong>Difficult to keep track of parameters</strong> we tweak in ML models. Small changes can lead to enormous differences in the results.</li><li>We have to<strong> keep track of the features the model works</strong> with, feature engineering is a separate task that contributes largely to model accuracy.</li><li><strong>Monitoring an ML model</strong> isn’t like monitoring a deployed software or web app.</li><li><strong>Debugging an ML model </strong>is an extremely complicated art<strong>.</strong></li><li>Models rely on real-world data for predicting, <strong>as real-world data changes, so should the model<em>.</em></strong> This means we have to keep track of new data changes and make sure the model learns accordingly.</li></ul><h3>How did we use MLOps Methodology?</h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2FkQ3FSVoJrkYWk%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2FkQ3FSVoJrkYWk%2Fgiphy.gif&amp;image=https%3A%2F%2Fi.giphy.com%2Fmedia%2FkQ3FSVoJrkYWk%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="244" frameborder="0" scrolling="no"><a href="https://medium.com/media/e15ecb286d0163f0f1dca1b4551b4984/href">https://medium.com/media/e15ecb286d0163f0f1dca1b4551b4984/href</a></iframe><p>We also had the same expression as above when we were are trying to convert our understanding of all the above MLOps learning into practice.</p><p>Our team explored and compared the pros and cons of different tools and services across the open-source tools and cloud-based managed services.</p><p>We not only emphasized choosing a unified MLOps Technology but also focused on having a platform where every team member in the organization whether he is an application developer, data analyst, or even business user gets the opportunity to explore the capabilities and benefits of augmenting their workflow or products with ML even if they don’t have the expertise required to create ML models.</p><p>The managed service of different platforms like GCP, AWS, and Azure were very fascinating as it had made the implementation of the MLOps methodology way much easier by removing the overhead complexity of</p><ul><li>Infrastructure Setup</li><li>Management</li><li>Maintenance</li><li>Scalability</li><li>Security</li><li>Integration</li></ul><p>After comparing different existing Cloud-based Services tools and taking into consideration of our existing Cloud applications we had decided to adopt the <strong><em>GCP Vertex AI as a one-stop platform for MLOps.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/864/1*uo6DA4bHHciVR44A7fh5jA.png" /><figcaption>Credit <a href="https://medium.com/@vineetjaiswal/introduction-comparison-of-mlops-platforms-aws-sagemaker-azure-machine-learning-gcp-vertex-ai-9c1153399c8e">Vineet Jaiswal</a></figcaption></figure><h3><strong>What is Vertex AI ?</strong></h3><blockquote>Vertex AI brings together the Google Cloud services for building ML under one, unified UI and API. In Vertex AI, you can now easily train and compare models using <a href="https://cloud.google.com/vertex-ai/docs/beginner/beginners-guide/">AutoML</a> or custom code training and all your models are stored in one central model repository. These models can now be deployed to the same endpoints on Vertex AI.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/761/1*SEHvg7507WDXhR5qGjh-GQ.png" /><figcaption>Vertex AI provides a unified set of APIs for the ML lifecycle</figcaption></figure><p><a href="https://cloud.google.com/ai-platform-unified/docs?utm_source=codelabs&amp;utm_medium=et&amp;utm_campaign=CDR_sar_aiml_vertexio_&amp;utm_content=-">Vertex AI</a> integrates the ML offerings across Google Cloud into a seamless development experience. Previously, models trained with AutoML and custom models were accessible via separate services. The new offering combines both into a single API, along with other new products. You can also migrate existing projects to Vertex AI.</p><p>In addition to model training and deployment services, Vertex AI also includes a variety of MLOps products, including Vertex Pipelines (the focus of this lab), Model Monitoring, Feature Store, and more. You can see all Vertex AI product offerings in the diagram below.</p><figure><img alt="Credit: GCP Vertex AI Docs" src="https://cdn-images-1.medium.com/max/778/1*8AbT7lvDB6Wjpqw4F6JFpw.png" /><figcaption>Credit: GCP Vertex AI Docs</figcaption></figure><h3>Conclusion</h3><p>Phew! That was quite a lot of Theoretical Knowledge Adventure. Give yourself a pat on your back. 👋</p><p>We covered</p><ul><li>What is MLOps?</li><li>Why does it matter?</li><li>Why did Zeals need to adopt it?</li><li>Why did we adopt the GCP Vertex AI Platform for MLOps and its overview?</li></ul><blockquote><em>We will try to bring in practical use cases in our upcoming blogs to show you how our company is adopting and utilizing MLOps methodology with GCP Vertex AI</em>.</blockquote><p>Stay tuned for that next piece of blog! Thank you for reading, and stay safe!</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2FQJE93F8fx3Bzjc62zJ%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2FQJE93F8fx3Bzjc62zJ%2Fgiphy.gif&amp;image=https%3A%2F%2Fi.giphy.com%2Fmedia%2FQJE93F8fx3Bzjc62zJ%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="183" frameborder="0" scrolling="no"><a href="https://medium.com/media/f555428c8602c4a580da2c64bde86bcc/href">https://medium.com/media/f555428c8602c4a580da2c64bde86bcc/href</a></iframe><h3>Reference:</h3><ul><li><a href="https://medium.com/@vineetjaiswal/introduction-comparison-of-mlops-platforms-aws-sagemaker-azure-machine-learning-gcp-vertex-ai-9c1153399c8e">Comparison of Cloud-based MLOps platforms</a> by <a href="https://medium.com/@vineetjaiswal?source=post_page-----9c1153399c8e-----------------------------------">Vineet Jaiswal</a> — Awesome blog recently published with the same idea and comparison stats as Zeals. (It came out when I was drafting this blog post. He has done justice to it and so we did not want to recreate it and used the same comparison table)</li><li><a href="https://globalcloudplatforms.com/2021/08/13/new-to-ml-learning-path-on-vertex-ai/">New To ML: Learning Path On Vertex AI</a></li><li><a href="https://towardsdatascience.com/a-gentle-introduction-to-mlops-7d64a3e890ff">A Gentle Introduction to MLOps</a> by <a href="https://medium.com/@yashaswi_nayak">Yashaswi Nayak</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a88612762895" width="1" height="1" alt=""><hr><p><a href="https://medium.com/zeals-tech-blog/adopting-mlops-methodology-in-zeals-a88612762895">Adopting MLOps Methodology in Zeals</a> was originally published in <a href="https://medium.com/zeals-tech-blog">The Zeals Tech Blog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>