<?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 Shorabh on Medium]]></title>
        <description><![CDATA[Stories by Shorabh on Medium]]></description>
        <link>https://medium.com/@code4545454545?source=rss-4a6e8fa3a60c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*6_uhk2GDcjlpt2nMsA25qQ.jpeg</url>
            <title>Stories by Shorabh on Medium</title>
            <link>https://medium.com/@code4545454545?source=rss-4a6e8fa3a60c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 15:38:54 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@code4545454545/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[I tested Temple for a 10 km run to confirm if its Brain Flow Readings Mean Anything]]></title>
            <link>https://medium.com/@code4545454545/i-bought-temple-to-see-whether-its-brain-flow-readings-mean-anything-ecf7e593229d?source=rss-4a6e8fa3a60c------2</link>
            <guid isPermaLink="false">https://medium.com/p/ecf7e593229d</guid>
            <category><![CDATA[temple]]></category>
            <category><![CDATA[fitness]]></category>
            <category><![CDATA[wearables]]></category>
            <category><![CDATA[biohacking]]></category>
            <dc:creator><![CDATA[Shorabh]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 10:14:13 GMT</pubDate>
            <atom:updated>2026-11-13T10:16:14.595Z</atom:updated>
            <content:encoded><![CDATA[<p>Testing Temple’s “Brain Flow” Metric: Claims, Evidence, and a Test Protocol</p><p>Temple is a head-worn wearable priced at ₹85,000 that reports a metric called Brain Flow. The company describes it as a proxy for cerebral blood flow. This post covers what is claimed, what has been validated, what critics dispute, and how I’m testing it. I make no claim about whether it works.</p><p><a href="https://www.business-standard.com/india-news/what-is-the-controversy-over-deepinder-goyal-s-85-000-temple-wearable-126111000259_1.html">business-standard</a><a href="https://www.business-standard.com/technology/tech-news/temple-explained-deepinder-goyal-s-bet-on-wearables-to-extend-healthy-life-126100500782_1.html">business-standard</a></p><p><strong>1. What is claimed</strong><br> Continuous, wearable-based tracking of changes in <strong>cerebral blood flow</strong> during daily activity. The founder has also said it is a fitness and AI product, not a medical device, and that Brain Flow is one of two metrics rather than the main focus. <a href="https://www.business-standard.com/india-news/what-is-the-controversy-over-deepinder-goyal-s-85-000-temple-wearable-126111000259_1.html">business-standard</a></p><p><strong>2. Evidence so far</strong><br> One validation study. It compared the metric with transcranial Doppler (TCD) measurements of middle cerebral <strong>artery blood-flow velocity</strong> in 20 participants, during stationary cycling and a standing-to-lying transition, with a median within-subject correlation of about 0.8. <a href="https://inc42.com/buzz/deepinder-goyals-temple-takes-first-step-towards-validating-its-wearable/">inc42</a></p><p>What this supports: the metric moves in the same direction as MCA velocity under controlled, short-term conditions. What it doesn’t support: absolute cerebral blood flow measurement or diagnosis. Two further limits worth noting: <a href="https://inc42.com/buzz/deepinder-goyals-temple-takes-first-step-towards-validating-its-wearable/">inc42</a></p><ul><li>Correlation measures co-movement, not agreement. A Bland-Altman analysis or error bounds would say more about accuracy.</li><li>A median across participants hides the spread. Per-subject results would show how consistent it is.</li></ul><p><strong>3. The objection</strong><br> Critics argue the device can possibly detect extracranial flow changes but not reliably blood flow to the brain. This is a signal-source problem. A sensor at the temple sits over scalp and superficial vessels, which respond to heat, exertion and posture independently of cerebral perfusion. Goyal’s counter is that detecting skin-level changes doesn’t establish that the metric measures blood flow inside the brain. I’m treating that as unresolved. <a href="https://www.freepressjournal.in/tech/deepinder-goyals-85000-temple-wearable-faces-doctors-backlash-over-brain-blood-flow-monitoring-claims-lack-of-scientific-evidence">freepressjournal</a><a href="https://www.business-standard.com/india-news/what-is-the-controversy-over-deepinder-goyal-s-85-000-temple-wearable-126111000259_1.html">business-standard</a></p><p><strong>4. Hypotheses</strong></p><ul><li>H1: Brain Flow readings are repeatable across identical sessions.</li><li>H2: Brain Flow responds to manipulations that should change cerebral perfusion but not skin perfusion in the same way.</li><li>H3: Brain Flow is not simply a restatement of heart rate or skin temperature.</li></ul><p><strong>5. Protocol</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/1*k6VKLaU3FU02EE3bsCagnQ.png" /><figcaption>Temple sensor placement at the side of the head, photographed [right after a 10 km run]. Sweat and motion are possible confounds for the signal, so I’m logging placement and conditions each session.</figcaption></figure><ul><li>Controls: resting baselines on separate days; posture changes (sitting, standing, lying); paced breathing at fixed rates.</li><li>Load: 100 pushups and a 10 km run, logged with a chest-strap heart rate for comparison.</li><li>Log per session: time of day, temperature, sensor placement, sweat, sleep, caffeine.</li><li>Analysis: compare Brain Flow against heart rate and against my own effort scores, and check run-to-run variance.</li></ul><p><strong>6. Limitations</strong><br> n = 1, no reference measurement (no TCD), no blinding, and exercise adds motion and sweat artifacts. Results can show consistency or inconsistency. They cannot show that the metric measures brain blood flow.</p><p><strong>7. Results</strong><br> [Not yet delivered]</p><p><strong>Open questions</strong></p><ul><li>Has anyone published the sensing method and signal-processing pipeline?</li><li>Are there independent replications beyond the first study?</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ecf7e593229d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Deloitte India Interview Experience | Generative AI / Agentic AI Developer (Consultant Role)]]></title>
            <link>https://medium.com/@code4545454545/deloitte-india-interview-experience-generative-ai-agentic-ai-developer-consultant-role-826695f0043e?source=rss-4a6e8fa3a60c------2</link>
            <guid isPermaLink="false">https://medium.com/p/826695f0043e</guid>
            <category><![CDATA[rags]]></category>
            <category><![CDATA[ai-engineering]]></category>
            <category><![CDATA[genai]]></category>
            <category><![CDATA[interview-preparation]]></category>
            <category><![CDATA[deloitte]]></category>
            <dc:creator><![CDATA[Shorabh]]></dc:creator>
            <pubDate>Thu, 12 Nov 2026 09:37:39 GMT</pubDate>
            <atom:updated>2026-11-12T09:37:39.542Z</atom:updated>
            <content:encoded><![CDATA[<p>I recently attended a <strong>Deloitte Walk-in Interview</strong>, and I wanted to share my interview experience. Hopefully this helps others preparing for similar roles.</p><p><strong>Round 1: Group Discussion (GD)</strong><br>Topic: <em>“Staying calm is more dangerous than arguing in meetings.”</em></p><p>The discussion focused on balancing emotional intelligence with constructive debate in professional environments.</p><h3>Technical Round 1 (35 minutes)</h3><p>The first technical round mainly focused on my <strong>projects and GenAI concepts</strong>.</p><p><strong>Questions asked:</strong></p><ol><li>Self introduction</li><li>Explain your project</li><li>Follow-up questions related to the project</li><li>What tech stack did you use to build the project?</li><li>Since LangChain is open source, how do you handle security concerns?</li><li>What is <strong>MCP</strong> and why is it used?</li><li>What is <strong>LangGraph</strong> and when should it be used?</li><li>What is <strong>RAG (Retrieval-Augmented Generation)</strong> and what are its use cases?</li><li>Do you have experience using <strong>Microsoft Azure</strong> for building AI applications?</li><li>How did you deploy your project and what technologies were used?</li><li>What are your day-to-day activities at work?</li><li>From whom do you receive tasks or project assignments?</li><li>Who decides the architecture (e.g., LLMs, embeddings, etc.)?</li><li>Any questions for the interviewer?</li></ol><p><strong>Questions asked:</strong></p><ol><li>How is your day going?</li><li>How was your first round? (They mentioned they would adjust the questions based on my feedback)</li><li>Self introduction</li><li>Tell me about the projects you have worked on</li><li>What is an <strong>AI Agent</strong>?</li><li>How do agents communicate with each other and perform reasoning?</li><li>If the retrieval quality in RAG is poor, what improvements would you make?</li><li>Suppose you build a <strong>RAG system connected to a database</strong>, where normal users should not access certain data but architects can access everything. How would you implement this? <em>(RBAC scenario)</em></li><li>How would you process data that is <strong>not in text format</strong> (images, tables, etc.)?</li><li>How would you implement <strong>security for FastAPI endpoints</strong>?</li><li>What deployment method did you use? (Follow-up: How does Docker work?)</li><li>How would you ensure <strong>security for data retrieved from RAG</strong>?</li><li>Any questions for the interviewer?</li><li>Current CTC, Expected CTC, Notice period.</li></ol><p>Overall, the interview focused heavily on <strong>GenAI architecture, RAG systems, AI agents, deployment, and security practices</strong>.</p><p>Sharing this to help anyone preparing for <strong>AI/GenAI roles</strong>. Hope this gives some idea of what to expect.</p><p>#InterviewExperience #Deloitte #GenAI #RAG #LangChain #AIEngineering #InterviewPreparation</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=826695f0043e" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>