<?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 Prashant Mhatre on Medium]]></title>
        <description><![CDATA[Stories by Prashant Mhatre on Medium]]></description>
        <link>https://medium.com/@mhatrep?source=rss-4ab625b9cc87------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*nFtttB1DwBrQ1D06XbbQPw.png</url>
            <title>Stories by Prashant Mhatre on Medium</title>
            <link>https://medium.com/@mhatrep?source=rss-4ab625b9cc87------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 12:46:19 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mhatrep/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[Understanding Metrics]]></title>
            <link>https://medium.com/@mhatrep/understanding-metrics-79057fa4c3f8?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/79057fa4c3f8</guid>
            <category><![CDATA[data-modeling]]></category>
            <category><![CDATA[metrics]]></category>
            <category><![CDATA[business-intelligence]]></category>
            <category><![CDATA[semantic-layer]]></category>
            <category><![CDATA[data-architecture]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Sun, 13 Sep 2026 03:28:12 GMT</pubDate>
            <atom:updated>2026-09-13T03:28:12.016Z</atom:updated>
            <content:encoded><![CDATA[<p>A metric is not just a number. A properly defined enterprise metric should have:</p><blockquote>Metric = Definition + Formula + Grain + Dimensions + Time + Aggregation + Unit + Target + Direction + Data Source + Owner</blockquote><p>For example:</p><blockquote><strong>Revenue</strong> = SUM(order_amount) at <strong>order grain</strong>, by <strong>customer/product/geography/time</strong>, measured in <strong>USD</strong>, with defined inclusion/exclusion rules.</blockquote><p>That distinction becomes extremely important when moving from <strong>dimensional modeling → semantic models → metric views → BI → Genie/agents</strong>, because two teams can both say “Revenue” while actually calculating two different metrics.</p><p>If you’re designing an enterprise <strong>semantic layer / metric store</strong>, think:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/483/1*pGD6QhGqbfguJjwiKvdRBw.png" /><figcaption>Metrics — The Ultimate Mental Model</figcaption></figure><h4>The Most Important Metric Families</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/811/1*fbCK0N1SRFDcdkx0Vs-ZzA.png" /><figcaption>Most Important Metric Families</figcaption></figure><h4>Don’t Mix These Three Dimensions</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/711/1*atgsBN1cBzI_0yPYYyhT5g.png" /><figcaption>Three Dimensions of Metrics</figcaption></figure><p>A metric is a defined way of measuring reality. The key is to understand its type, purpose, time dimension, behavior, and target so that metrics become trusted tools for decisions rather than just numbers on a dashboard.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=79057fa4c3f8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Six Ingradients of an Effective Prompt for AI Engineer]]></title>
            <link>https://medium.com/@mhatrep/six-ingradients-of-an-effective-prompt-for-ai-engineer-7faeb957d799?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/7faeb957d799</guid>
            <category><![CDATA[context-engineering]]></category>
            <category><![CDATA[writing-prompts]]></category>
            <category><![CDATA[prompt-engineering]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[generative-ai-tools]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Fri, 24 Jul 2026 04:52:04 GMT</pubDate>
            <atom:updated>2026-07-24T04:52:04.809Z</atom:updated>
            <content:encoded><![CDATA[<p>— — — — — — — — — — — — — — — — — — —</p><p>Regardless of tool, every strong prompt is built from the same six ingredients. Think of it as a checklist, not a template you fill in word‐for‐word.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QwIqMdqyI0EwW9y20CgLiQ.png" /><figcaption>Anatomy of an Effective Prompt</figcaption></figure><h4>Six Ingradients of an Effective Prompt</h4><p>— — — — — — — — — — — — — — — — — — — — — -</p><p><strong>1. Role / Persona:</strong> Role / Persona Anchors tone, expertise level, and the kind of judgment calls the model should make</p><blockquote>You are a senior backend engineer who prioritizes readability and test coverage over cleverness.</blockquote><p><strong>2. Context: </strong>Context The facts the model can’t infer i. e. files, schemas, error logs, business rules</p><blockquote>Here is the current schema: &lt;paste DDL&gt;. This service handles payment webhooks.</blockquote><p><strong>3. Task / Instruction :</strong> One specific, actionable ask, not a vague direction</p><blockquote>Add input validation to &lt;function_name&gt; that rejects negative amounts.</blockquote><p><strong>4 . Constraints Guardrails :</strong> style, scope, what not to touch</p><blockquote>Don’t change the public API signature. Don’t add new dependencies.</blockquote><p><strong>5. Examples (few‐shot) : </strong>1–3 samples of the input/output shape you want, especially for formatting‐sensitive tasks</p><blockquote>Format each finding like: [SEVERITY] file:line — — issue — — fix.</blockquote><p><strong>6. Output format : </strong>The schema or shape of the response- diff, table, JSON, checklist</p><blockquote>Return only a unified diff, no prose.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7faeb957d799" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mental Models for Behavioral Interviews]]></title>
            <link>https://medium.com/@mhatrep/mental-models-for-behavioral-interviews-37a92ab7357f?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/37a92ab7357f</guid>
            <category><![CDATA[technical-interview]]></category>
            <category><![CDATA[interview]]></category>
            <category><![CDATA[job-interview]]></category>
            <category><![CDATA[behavioral-interviews]]></category>
            <category><![CDATA[job-hunting]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Wed, 15 Jul 2026 01:59:16 GMT</pubDate>
            <atom:updated>2026-07-15T01:59:16.806Z</atom:updated>
            <content:encoded><![CDATA[<p>A good behavioral interview answer is about<em> proving a competency.</em> Every question (“Tell me about a conflict”, “Describe a failure”, “Leadership example”) is really asking: <em>“Can you demonstrate evidence that you possess a competency we need?”</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*btJv72IiydTdirVQzEaOBA.png" /><figcaption>Interviewer’s Evaluation Model</figcaption></figure><h3>🤺 <strong>Universal </strong>Behavioral Interviews <strong>Mental Model</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/244/1*Pk9JFvsZEFIDBcZ-FbAqPQ.png" /></figure><p>Think:</p><blockquote><strong>Competency → Evidence → Results</strong></blockquote><h3>🤺 <strong>The 6-Step Framework for </strong>Behavioral Interviews</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Al4tc3355RX5E1523ILwrw.png" /></figure><p><strong>Every Behavioral Question Tests One of These</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/919/1*X-e0NxW3QEJ5ApktF5DLEA.png" /></figure><p>Nearly every behavioral interview question fits one or more of these categories.</p><h3>🤺<strong>Story Structure for </strong>Behavioral Interview</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/802/1*GZd-otZfZmvagdvt1TumwA.png" /></figure><p>This is an enhanced version of the common STAR method.</p><h3>🤺<strong>Interviewer’s Mental Checklist</strong></h3><p>While you speak, the interviewer is silently asking:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8B9nNN-_Aht2TyWsC8MpJQ.png" /><figcaption>Interviewer’s Mental Checklist</figcaption></figure><p>Answer these implicitly through your story.</p><h3>🤺<strong>Think Like an Architect</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/148/1*q92pq2vnB11nmHIHLbDrtA.png" /><figcaption>Think Like an Architect</figcaption></figure><p>This decision-making flow works well for architecture, consulting, and leadership roles.</p><h3>🤺<strong>Story Inventory Matrix</strong></h3><p>Prepare stories that can be reused across multiple questions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CeCTayUOAm8CWI7H5yL4Vw.png" /><figcaption>Story Inventory Matrix</figcaption></figure><p>With 8–10 strong stories, you can answer dozens of behavioral questions.</p><h3>🍁<strong>Action Formula</strong></h3><p>Instead of saying:</p><blockquote>“We fixed the issue.”</blockquote><p>Say:</p><blockquote>I analyzed<br>I identified<br>I proposed<br>I convinced<br>I coordinated<br>I designed<br>I implemented<br>I monitored<br>I optimized<br>I documented</blockquote><p>These action verbs emphasize your individual contribution.</p><h3>🍁<strong>Results Formula</strong></h3><p>Good results combine measurable outcomes with business value.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QVK-BtO000uQwZ9Fg0NDmg.png" /><figcaption>Results Formula</figcaption></figure><p>Whenever possible, quantify your impact.</p><h3>🍁<strong>Reflection Formula</strong></h3><p>Every story should end with:</p><ol><li>What worked?</li><li>What would you do differently?</li><li>How has this changed how you work today?</li></ol><p>Example:</p><blockquote>“That experience taught me to involve stakeholders earlier and validate assumptions before implementation. Since then, I consistently schedule early design reviews, which has helped prevent similar issues.”</blockquote><h3>🤺<strong>The CARE Mental Model</strong></h3><p>A simple mnemonic to remember during interviews:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3UCUKndnaMZDODXB7Vjc9g.png" /><figcaption>CARE Mental Model</figcaption></figure><h3>🤺<strong>Quick Mapping: Question → Competency</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kCVLy9eu9o9_h32SeCyFPA.png" /><figcaption>Competency Mapping</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1009/1*D61XQ-pcEY5G1i26RILlIg.png" /></figure><h3>🤺<strong>The Ultimate Mental Model</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jqCYWZOF2AfE1PS-IJJusA.png" /></figure><p>If you consistently follow this model, you’ll answer behavioral questions in a structured, credible way while highlighting the qualities interviewers are evaluating.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=37a92ab7357f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Data Architect Mental Model]]></title>
            <link>https://medium.com/@mhatrep/data-architect-mental-model-26825635ec12?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/26825635ec12</guid>
            <category><![CDATA[data-architecture]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[data-modeling]]></category>
            <category><![CDATA[data-quality]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Sun, 21 Jun 2026 00:44:11 GMT</pubDate>
            <atom:updated>2026-06-21T00:44:11.710Z</atom:updated>
            <content:encoded><![CDATA[<p>Whenever you see a dataset, table, API, report, semantic model, ontology, AI agent, or pipeline, ask:</p><pre>Purpose    → Why?<br>Value      → What decision?<br>Reality    → What happened?<br>Meaning    → What does it mean?<br>Trust      → Can I trust it?<br>Control    → Who owns/governs it?<br>Structure  → How is it organized?<br>Flow       → How does it move?<br>Evolution  → How will it change?</pre><h3>Complete Data Architect Mental Model</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RirREzM2RO7fBi2NaQHhIg.png" /></figure><h3>The 50 Things Expert Architects Constantly Think About</h3><h4>Purpose</h4><ul><li>Strategy, Goals, KPIs, Regulations</li></ul><h4>Value</h4><ul><li>Decisions, Actions, Outcomes, ROI, Insights, Knowledge</li></ul><h4>Reality</h4><ul><li>Entities, Events, States, Event Time, Processing Time, Effective Time, Versions, Lifecycle</li></ul><h4>Meaning</h4><ul><li>Definitions, Metrics, Taxonomy, Ontology, Rules, Policies, Reference Data. Master Data</li></ul><h4>Trust</h4><ul><li>Accuracy, Completeness, Consistency, Timeliness, Lineage. Provenance, Auditability, Observability</li></ul><h4>Control</h4><ul><li>Ownership, Stewardship, Security, Privacy, Compliance, Access Controls, Retention</li></ul><h4>Structure</h4><ul><li>Relationships, Hierarchies, Granularity, Schemas, Models, Graphs, Documents, Vectors, Keys, Partitions</li></ul><h4>Flow</h4><ul><li>Sources, Ingestion, ETL, ELT, CDC, Streaming, APIs, Processing, Consumption</li></ul><h4>Evolution</h4><ul><li>Versioning, Scalability, Extensibility, Change Management, Schema Evolution, Rule Evolution, Metric Evolution</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=26825635ec12" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Taxonomy of Business Problems for AI]]></title>
            <link>https://medium.com/@mhatrep/taxonomy-of-business-problems-for-ai-6cf96e6f2b3e?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/6cf96e6f2b3e</guid>
            <category><![CDATA[analytics]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[business-architecture]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Sat, 20 Jun 2026 03:16:25 GMT</pubDate>
            <atom:updated>2026-06-20T03:16:25.000Z</atom:updated>
            <content:encoded><![CDATA[<p>Almost every business problem that ends up on a data team’s desk falls into a small number of underlying shapes, regardless of the language used to describe it.</p><p>Recognizing the shape is most of the work; once youknow the shape, the candidate techniques narrow dramatically.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qtK04vxgJk0hciT-YGG9XQ.png" /><figcaption>Problem Shape, Question, Technique</figcaption></figure><p>The <strong>eight shapes</strong> that cover the overwhelming majority of real requests are:</p><ol><li>Predicting a number or category for a new case</li><li>Finding similar or relevant items</li><li>Generating new content</li><li>Optimizing a decision under constraints</li><li>Understanding cause and effect</li><li>Detecting anomalies</li><li>Grouping similar things without predefined labels</li><li>Enforcing consistent rules or policy</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/644/1*APIgCpcgiyq4TNEU2Hw1zQ.png" /><figcaption>The eight recurring shapes behind most business requests for AI or analytics.</figcaption></figure><p>Most ‘we need AI for X’ requests are really one of these eight wearing business-specific clothing.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6cf96e6f2b3e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building Responsible AI with Amazon Bedrock Guardrails]]></title>
            <link>https://medium.com/@mhatrep/building-responsible-ai-with-amazon-bedrock-guardrails-26ac672fded2?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/26ac672fded2</guid>
            <category><![CDATA[aws-bedrock]]></category>
            <category><![CDATA[ai-compliance]]></category>
            <category><![CDATA[responsible-ai]]></category>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[aws-bedrock-guardrails]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Sun, 23 Feb 2025 18:25:09 GMT</pubDate>
            <atom:updated>2025-02-23T18:25:09.011Z</atom:updated>
            <content:encoded><![CDATA[<p>As generative AI becomes more integral to applications, ensuring safety, accuracy, and compliance is crucial. Amazon Bedrock Guardrails provides configurable safeguards to help organizations build responsible AI applications at scale. With industry-leading safety protections, Guardrails ensures factual accuracy, blocks harmful content, and protects user privacy while maintaining a seamless AI experience.</p><p><a href="https://aws.amazon.com/bedrock/">https://aws.amazon.com/bedrock/</a></p><p>This article explores the capabilities of <strong>Amazon Bedrock Guardrails</strong>, how it helps mitigate AI risks, and how you can integrate it into your applications.</p><h4>What is Amazon Bedrock Guardrails?</h4><p>Amazon Bedrock Guardrails is a <strong>responsible AI solution</strong> designed to <strong>enhance safety, privacy, and truthfulness</strong> in generative AI applications. It applies a consistent and standardized approach across various <strong>foundation models (FMs)</strong>, fine-tuned models, and <strong>self-hosted AI models</strong>.</p><h4>Key Features of Amazon Bedrock Guardrails</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uJxNzav7BAfgvfssYvuohQ.png" /><figcaption>Amazon Bedrock Guardrails — Key Features Summary</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Puu-STAqaCreaCraoi83yg.png" /><figcaption>Amazon Bedrock Guardrails — Key Features</figcaption></figure><h3>How Amazon Bedrock Guardrails Helps Ensure Responsible AI</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sSqQbMu5AtYmdEmZ53B18Q.png" /><figcaption>How Amazon Bedrock Guardrails Helps Ensure Responsible AI</figcaption></figure><h4>1. Detect &amp; Prevent Hallucinations</h4><p>AI models sometimes <strong>generate incorrect or misleading responses</strong> due to hallucinations. Guardrails introduces <strong>contextual grounding checks</strong>, ensuring AI-generated responses are <strong>factually accurate and relevant</strong>.</p><blockquote>🦖 <strong>Example Use Case</strong>: A <strong>customer support chatbot</strong> using RAG validates its responses against a knowledge base to avoid making up answers.</blockquote><h4>2. Automated Reasoning for Fact-Checking</h4><p>Guardrails’ <strong>Automated Reasoning Checks</strong> (Preview) use <strong>formal logic</strong> to verify AI responses against known facts. Organizations can upload HR guidelines, compliance manuals, or operational handbooks, and Bedrock generates policies to validate AI-generated content.</p><blockquote>🦖 <strong>Example Use Case</strong>: A <strong>financial institution</strong> uses Automated Reasoning to ensure compliance with <strong>investment regulations</strong>, preventing AI from offering unauthorized financial advice.</blockquote><h4>3. Block Harmful &amp; Undesirable Content</h4><p>Guardrails <strong>filters and blocks</strong> potentially <strong>toxic or unsafe</strong> content, including:</p><ul><li>Hate speech</li><li>Insults</li><li>Sexual content</li><li>Violence</li><li>Criminal activity</li><li>Prompt injections &amp; jailbreaks</li></ul><blockquote>🦖 <strong>Example Use Case</strong>: An <strong>AI-powered travel assistant</strong> blocks discussions about political or financial topics, ensuring <strong>brand safety and compliance</strong>.</blockquote><h4>4. Protect Privacy with PII Redaction</h4><p>To prevent AI from exposing <strong>sensitive user data</strong>, Guardrails detects and redacts <strong>PII (names, emails, phone numbers, etc.)</strong>. You can define <strong>custom rules using RegEx</strong> for additional control.</p><blockquote>🦖 <strong>Example Use Case</strong>: A <strong>call center AI</strong> automatically <strong>redacts personal details</strong> while summarizing customer interactions.</blockquote><h4>5. Govern AI Responses Across Applications</h4><p>With <strong>ApplyGuardrail API</strong>, developers can <strong>evaluate AI-generated responses</strong> from <strong>third-party</strong> and <strong>self-managed models</strong>. This ensures <strong>consistent safety policies</strong> across all AI applications.</p><blockquote>🦖 <strong>Example Use Case</strong>: A company using <strong>Meta Llama on Amazon SageMaker</strong> applies <strong>Guardrails API</strong> to monitor and block inappropriate responses before presenting them to users.</blockquote><h3>Step-by-Step Guide: Setting Up Amazon Bedrock Guardrails</h3><h4>1️⃣ Create a Guardrail in AWS Console</h4><ol><li>Navigate to Amazon Bedrock → Guardrails</li><li>Click “Create Guardrail”</li><li>Define safety settings for:</li></ol><ul><li>Denied Topics</li><li>Content Filtering</li><li>PII Redaction</li><li>Automated Reasoning (Preview)</li></ul><p>4. Save &amp; Deploy the Guardrail</p><h4>2️⃣ Test Guardrails with a Model</h4><ul><li>Select a <strong>foundation model</strong> (e.g., Anthropic Claude, AI21 Jurassic, Amazon Titan).</li><li>Enter a <strong>prompt</strong> and <strong>review the Guardrail intervention logs</strong>.</li></ul><h4>3️⃣ Use ApplyGuardrail API for Custom Models</h4><p>For independent model validation, use the <strong>ApplyGuardrail API</strong> to check inputs and outputs across AI applications.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MV8AhnsathhLwUvUyiq5ow.png" /><figcaption>Python Sample</figcaption></figure><h4><strong>Real-World Use Cases</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NjzEf7720MNXAfp0RjL7kw.png" /><figcaption><strong>Responsible AI — Real-World Use Cases</strong></figcaption></figure><blockquote>Amazon Bedrock Guardrails provides <strong>an essential safety layer</strong> for building <strong>responsible, secure, and privacy-compliant</strong> generative AI applications. With powerful features like <strong>hallucination detection, Automated Reasoning, content filtering, and PII redaction</strong>, organizations can confidently <strong>deploy AI while ensuring safety and compliance</strong>.</blockquote><p>Reference: <a href="https://aws.amazon.com/bedrock/">https://aws.amazon.com/bedrock/</a> and blog articles mentioned on the page.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=26ac672fded2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Machine Learning (ML) Model Building and Testing — Hello World!]]></title>
            <link>https://medium.com/@mhatrep/machine-learning-ml-model-building-and-testing-hello-world-08496293ab86?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/08496293ab86</guid>
            <category><![CDATA[machine-learning-ai]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[model-building]]></category>
            <category><![CDATA[model-serving]]></category>
            <category><![CDATA[python]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Sat, 16 Nov 2024 19:02:46 GMT</pubDate>
            <atom:updated>2024-11-16T19:02:46.805Z</atom:updated>
            <content:encoded><![CDATA[<h3>Machine Learning (ML) Model Building and Testing — Hello World!</h3><p><strong>Example Use case:</strong> Develop a machine learning model that predicts house prices based on the square footage of a property.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/988/1*T-LT94M8eeKVcDnFTDU5cA.png" /><figcaption>ML Model Building and Serving</figcaption></figure><p>I want to show simplistic process to start from data, build and serve the model. This does <strong>NOT </strong>teach you machine learning but explains building and serving model in a simplistic way. (For the people who know Python, and basics of ML but get lost in tons of documentation)</p><ol><li>(Optional) Generate Data (or Use your data)</li><li>Build and Save Model</li><li>Serve Model</li></ol><blockquote>&gt; pip install pandas numpy scikit-learn flask joblib</blockquote><h3>1. ✱<strong>Generate Data</strong>✱</h3><blockquote><a href="https://github.com/mhatrep/simple_ml_example/blob/main/1_data_generator.py">https://github.com/mhatrep/simple_ml_example/blob/main/1_data_generator.py</a></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/311/1*qufRaphii4VB-HKfYdBaEA.png" /><figcaption>Generate Synthetic Data</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/852/1*ZsWZCdGSu9cTSZMtq4ksag.png" /><figcaption><a href="https://github.com/mhatrep/simple_ml_example/blob/main/1_data_generator.py">1_data_generator.py</a></figcaption></figure><p>☝<strong><em>&gt; python 1_data_generator.py<br></em></strong>This will create <strong>data.csv</strong> file</p><h3>2. ✱Build and Save Model✱</h3><blockquote><a href="https://github.com/mhatrep/simple_ml_example/blob/main/2_train_build_model.py">https://github.com/mhatrep/simple_ml_example/blob/main/2_train_build_model.py</a></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gh3VoWLrVNhSnv1KXnq_fA.png" /><figcaption><a href="https://github.com/mhatrep/simple_ml_example/blob/main/2_train_build_model.py">2_train_build_model.py</a></figcaption></figure><p>☝<strong><em>&gt; python 2_train_build_model.py<br></em></strong>This will create <strong>model.joblib</strong> file</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zNBT6DlwUHPBHUTaV7bUhw.png" /><figcaption>Model Format Options</figcaption></figure><h3>3. ✱Serve Model✱</h3><blockquote><a href="https://github.com/mhatrep/simple_ml_example/blob/main/3_app.py">https://github.com/mhatrep/simple_ml_example/blob/main/3_app.py</a></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vaOs8owjErwieGeQN8WwEg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1020/1*khoG5wOgIsEvAp1zMoKMjQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/991/1*5EXY3HEH-9Y91kxO6HWpmQ.png" /><figcaption><a href="https://github.com/mhatrep/simple_ml_example/blob/main/3_app.py">3_app.py</a></figcaption></figure><p>☝<strong><em>&gt; python </em>3_app.py</strong></p><p>This will serve the model</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AF5bEWRf7wH3TSaTP8_vEw.png" /><figcaption>ML Model Serving</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/495/1*na0W4vrXs8NYt-2KNtH6qw.png" /></figure><h3>✱<strong>Test from UI</strong>✱</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/482/1*q2RS_v1KGkXKCPg8JJDkSA.png" /></figure><p>Enter data and Click ‘Predict’</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/711/1*NyP2ghnKraQHyKvHz5Po1A.png" /><figcaption>Browser to Test ML Model</figcaption></figure><h3>✱<strong>Test from Command Line</strong>✱</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BVsr4Td23jEHD6cecHmnfA.png" /><figcaption>CURL to Test ML Model</figcaption></figure><p>That’s it. Thank you for reading.</p><blockquote><strong>Note: </strong>ChatGPT assisted code generation</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=08496293ab86" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Amazon DataZone Tutorial & Overview]]></title>
            <link>https://medium.com/@mhatrep/amazon-datazone-tutorial-overview-ff01cffddb8e?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/ff01cffddb8e</guid>
            <category><![CDATA[amazon-datazone]]></category>
            <category><![CDATA[data-governance-tools]]></category>
            <category><![CDATA[data-governance-software]]></category>
            <category><![CDATA[data-governance-framework]]></category>
            <category><![CDATA[data-governance]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Sun, 02 Jun 2024 15:35:36 GMT</pubDate>
            <atom:updated>2024-06-02T15:35:36.714Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Amazon DataZone</strong> is a comprehensive data management service designed to streamline the processes of cataloging, discovering, sharing, and governing data stored across various platforms, including AWS, on-premises, and third-party sources. It facilitates efficient data access and collaboration across organizational boundaries while ensuring robust governance and security.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qvlHNYlEpnMvQgN-sZNalw.png" /><figcaption>Amazon DataZone: <a href="https://aws.amazon.com/datazone/">https://aws.amazon.com/datazone/</a></figcaption></figure><blockquote>This article provides an overview of <strong>Amazon DataZone</strong>, highlighting its key features, use cases, architecture, and steps to get started. By leveraging Amazon DataZone, organizations can enhance their data management capabilities and drive data-driven insights with ease.</blockquote><h4>🟦 Amazon DataZone — <strong>Key Features and Components</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IU9Yo-G5vX9ZkvIqjZkyxQ.png" /><figcaption>Amazon DataZone — Key Features and Components</figcaption></figure><ol><li>🍎<strong>Unified Data Management Portal</strong> Amazon DataZone provides a web-based application that serves as a centralized portal for users to catalog, discover, access, analyze, and govern data. This portal allows for seamless collaboration among data engineers, data scientists, product managers, analysts, and business users.</li><li>🍎<strong>Fine-Grained Access Controls</strong> Administrators and data stewards can manage and govern access to data using detailed access controls. This ensures that data is accessible with the appropriate level of privileges and context, enhancing data security and compliance.</li><li>🍎<strong>Built-In Workflows</strong> Amazon DataZone includes built-in workflows for data consumers to request access to data and for data owners to approve these requests. This automated process streamlines data access while maintaining strict governance.</li><li>🍎<strong>Integration with Analytics Tools</strong> DataZone integrates with various AWS analytics tools like Amazon Redshift Query Editor and Amazon Athena, allowing users to consume data directly from the DataZone portal without needing to log into the AWS Management Console.</li><li>🍎<strong>Support for Business Glossaries</strong> The service includes a business data catalog that supports a business glossary, providing consistent definitions for business terms across the organization. This enhances data understanding and ensures consistent use of terminology.</li><li>🍎<strong>Data Projects and Environments</strong> DataZone projects are groupings based on business use cases, where users, data assets, and analytics tools are organized to facilitate collaboration. Within projects, environments provide the necessary infrastructure and access controls for data and analytics tools.</li></ol><h4>🟦 Amazon DataZone <strong>Use Cases</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bAiMyX-5Df9GVVNeVDFyVg.png" /><figcaption>Amazon DataZone Use Cases</figcaption></figure><ul><li>🍍<strong>Find Data in the Business Data Catalog</strong> Users can search for, share, and access cataloged data using business terms, regardless of whether the data is stored on AWS, on-premises, or with third-party providers.</li><li>🍍<strong>Streamline Workflow Processes</strong> Collaboration across teams is made efficient with self-service access to data and analytics tools, enhancing productivity and workflow efficiency.</li><li>🍍<strong>Simplify Analytics Access</strong> Users receive a personalized view to discover, prepare, transform, analyze, and visualize data using the web-based application, simplifying the analytics process.</li><li>🍍<strong>Administrate Access from One Place</strong> Centralized management of data access ensures compliance with organizational security regulations, making governance easier.</li></ul><h4>🟦 <strong>Amazon DataZone Features and Integrations</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VfpT3SZ0nIPUtdqx5wJTqA.png" /><figcaption>Amazon DataZone Features and Integrations</figcaption></figure><p><strong>Amazon DataZone</strong> supports programmatic interactions through APIs, AWS CloudFormation, AWS Command Line Interface (CLI), and AWS Cloud Development Kit (CDK). This flexibility allows for seamless integration with existing processes and automation of data management tasks.</p><h4>🟦 Getting Started with Amazon DataZone</h4><p>Refer the article below for details:</p><p><a href="https://aws.amazon.com/blogs/aws/amazon-datazone-now-generally-available-collaborate-on-data-projects-across-organizational-boundaries/">Amazon DataZone Now Generally Available - Collaborate on Data Projects across Organizational Boundaries | Amazon Web Services</a></p><p>Below is the high level flow:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eM8qqPauZbPU9D-UwORzzw.png" /><figcaption>Getting Started with Amazon DataZone</figcaption></figure><ol><li>🍓<strong>Create a Domain</strong> Begin by creating a domain that encompasses all core components like the business data catalog, projects, and environments. This can be done through the Amazon DataZone console.</li><li>🍓<strong>Create Projects and Environments</strong> Within the data portal, create projects that align with business use cases and set up environments that provide the necessary infrastructure for data and analytics tools.</li><li>🍓<strong>Publish Data</strong> Use environments to publish data assets into the Amazon DataZone catalog. This includes defining metadata and setting up data access controls.</li><li>🍓<strong>Subscribe to Data</strong> Data consumers can search for and request access to published data assets. Approved requests grant access to the data, which can be analyzed using integrated tools like Amazon Redshift and Amazon Athena.</li></ol><blockquote><strong>Amazon DataZone</strong> is a data management service that facilitates the <strong><em>cataloging, discovery, sharing, and governance of data</em></strong> across AWS, on-premises, and third-party sources. It provides a unified data management portal with built-in workflows and fine-grained access controls, enhancing data accessibility and collaboration. Users can seamlessly integrate with AWS analytics tools to derive data-driven insights efficiently.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ff01cffddb8e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Increase Font Size in Unreal Engine 5 | UE5 Scale Text]]></title>
            <link>https://medium.com/@mhatrep/how-to-increase-font-size-in-unreal-engine-5-ue5-scale-text-384c787a45a1?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/384c787a45a1</guid>
            <category><![CDATA[game-development]]></category>
            <category><![CDATA[unreal-engine-5]]></category>
            <category><![CDATA[ue5]]></category>
            <category><![CDATA[virtual-production]]></category>
            <category><![CDATA[unreal-engine]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Sat, 04 May 2024 03:38:09 GMT</pubDate>
            <atom:updated>2024-05-04T03:38:09.165Z</atom:updated>
            <content:encoded><![CDATA[<p>Tiny font in Unreal Engine 5?</p><p>&gt; From Main Windows, press <strong>Ctrl+Shift+W </strong>to open <strong>Widget Reflector.</strong></p><p>&gt; Update <strong>Application Scale</strong>: I use 1.35 for my montior.</p><p>&gt; Close Window.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FPP2dERt2YEeopolnE7Fbg.png" /><figcaption>How to Increase Font Size in Unreal Engine 5 | UE5 Scale Text</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=384c787a45a1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What is “Value Addition”? Define Value Addition.]]></title>
            <link>https://medium.com/@mhatrep/what-is-value-addition-define-value-addition-c142fefc24d3?source=rss-4ab625b9cc87------2</link>
            <guid isPermaLink="false">https://medium.com/p/c142fefc24d3</guid>
            <category><![CDATA[value-additions]]></category>
            <category><![CDATA[save-money]]></category>
            <category><![CDATA[services]]></category>
            <category><![CDATA[make-mone]]></category>
            <category><![CDATA[money]]></category>
            <dc:creator><![CDATA[Prashant Mhatre]]></dc:creator>
            <pubDate>Mon, 08 Apr 2024 03:18:15 GMT</pubDate>
            <atom:updated>2024-04-08T03:18:15.068Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>“Value addition”</strong> is a common phrase in business, especially when talking about products or services. When I ask people what it really means, they usually list things like making a product better, adding new features, improving how it looks or is packaged, adding technology, making it more personal, or meeting certain quality standards etc.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/822/1*9ERrLVUg9N4xIwLRgDO22w.png" /></figure><h4>🟩 What is Value Addition?</h4><blockquote>✔ ️To me, value addition means doing something for <strong>addressing pain points</strong> for the customers that help them <strong>either earn more money or save money.</strong></blockquote><blockquote>This includes making things easier or faster for them, improving their health or life in general, and helping them work better. Saving time or making life better can also lead to saving or making money.</blockquote><p>Also, it’s important to remember that sometimes your real customer might actually be the customer of the person you’re selling to <strong>(Customer’s customer).</strong> So, think about what those end users really need.</p><p><strong>Value addition</strong> is really important for businesses. It makes their products or services stand out from others, which can let them charge more and maybe make more profit.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c142fefc24d3" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>