Back to home

ARTICLES FIRST. IDEAS THAT TRAVEL.

Thinking in public.
Collected in one place.

3 original technical perspectives on this site, alongside 21 published LinkedIn articles and 130 original posts.

Evidence before confidence.

Answer-led perspectives on trustworthy AI. Documented platform capabilities, architectural judgment, and the limits of the evidence are kept distinct.

LINKEDIN ARTICLES & ORIGINAL POSTS

151 entries

Original writing only. Post ages are as of October 12, 2026.

ArticleLinkedIn3 min read

The GCP Enterprise Data Solution Architecture - How to Integrate different types of data sources to generate effective Business Intelligence? (opens in a new tab)

Connecting enterprise data sources to business intelligence on Google Cloud.

Read the original: The GCP Enterprise Data Solution Architecture - How to Integrate different types of data sources to generate effective Business Intelligence? (opens in a new tab)
PostLinkedIn4d ago

Confessions of a Corporate AI Agent: A Story of Infinite Compute and 47-Step Approvals (opens in a new tab)

I can process 400 pages of data in 9 seconds. šŸš€ But I’ve been waiting on an IT access ticket for 3 weeks. 😩 Welcome to the life of an Enterprise AI Agent. Day 1: The CEO introduces me as ā€œthe future of work.ā€ 🌟 Day 2: IT says I need my manager’s approval...

Read the original: Confessions of a Corporate AI Agent: A Story of Infinite Compute and 47-Step Approvals (opens in a new tab)
PostLinkedIn1mo ago

The next phase of Enterprise AI won't be won by the organization with the cheapest model. (opens in a new tab)

It will be won by the organization with the lowest cost per successful business outcome. Most AI cost discussions focus on infrastructure: ā˜ļø Serverless ⚔ Reserved Capacity 🧠 Open Models In reality, the biggest savings often come from architecture decisions:...

Read the original: The next phase of Enterprise AI won't be won by the organization with the cheapest model. (opens in a new tab)
PostLinkedIn1mo ago

Seven models. One platform. Built for production economics. (opens in a new tab)

At Build 2026, Microsoft AI shipped a full first-party model family into Microsoft Foundry - reasoning, coding, image, voice, and transcription. What makes this commercially interesting isn't the leaderboard. It's the cost curve.šŸ“‰ 🧠MAI-Thinking-1 - a sparse...

Read the original: Seven models. One platform. Built for production economics. (opens in a new tab)
PostLinkedIn1mo ago

We spent the last decade securing users and applications. The next decade will be about securing AI agents. (opens in a new tab)

As organizations move from AI experiments to deploying thousands of AI agents, identity becomes the new control plane. We already know how to govern people (Entra ID) and workloads (Managed Identities). The next challenge is governing AI agents with the same...

Read the original: We spent the last decade securing users and applications. The next decade will be about securing AI agents. (opens in a new tab)
PostLinkedIn2mo ago

What if an executive briefing could automatically choose the right AI model for every task? (opens in a new tab)

You can build your Executive Weekly Digest Agent in Microsoft Foundry that converts scattered enterprise information into concise, decision-ready insights. Powered by Model Router, it intelligently selects the appropriate model based on task complexity -...

Read the original: What if an executive briefing could automatically choose the right AI model for every task? (opens in a new tab)
PostLinkedIn2mo ago

Enterprise AI needs more than a powerful model - it needs a secure, scalable operating foundation. (opens in a new tab)

This architecture brings together: šŸ”¹ Microsoft Foundry for models, agents and the model harness šŸ”¹ Azure API Management AI Gateway for policy, token controls and governance šŸ”¹ Private data and services for enterprise grounding šŸ”¹ Security, compliance and...

Read the original: Enterprise AI needs more than a powerful model - it needs a secure, scalable operating foundation. (opens in a new tab)
PostLinkedIn2mo ago

Open Models + Model Harnesses: Building Enterprise AI with Microsoft Foundry (opens in a new tab)

Open and open-weight models provide flexibility - but a model endpoint alone is not an enterprise AI solution. A model/agent harness adds the operational layer around the model: 🧠 Reasoning and orchestration šŸ› ļø Tools and API actions šŸ’¾ Memory and...

Read the original: Open Models + Model Harnesses: Building Enterprise AI with Microsoft Foundry (opens in a new tab)
PostLinkedIn3mo ago

The AI Architecture Layer Nobody Is Talking About Enough: The AI Gateway (opens in a new tab)

As enterprises race to deploy AI at scale, many are realizing that building AI applications is the easy part. Controlling them is the hard part. In a world where AI consumption is measured in tokens, every prompt, completion, and agent interaction has a cost...

Read the original: The AI Architecture Layer Nobody Is Talking About Enough: The AI Gateway (opens in a new tab)
PostLinkedIn4mo ago

Fresh from #MicrosoftBuild 2026 šŸ‘€ - OpenClaw Windows Companion (opens in a new tab)

What’s really interesting about the Microsoft + OpenClaw direction isn’t just autonomous agents - it’s the focus on responsible and governed AI usage. Microsoft appears to be adding enterprise controls around OpenClaw-style agents with: - permission-aware...

Read the original: Fresh from #MicrosoftBuild 2026 šŸ‘€ - OpenClaw Windows Companion (opens in a new tab)
PostLinkedIn4mo ago

Microsoft Build 2026 is here and the era of AI Agents is officially mainstream. (opens in a new tab)

Explore the event šŸ‘‰ https://lnkd.in/eajJYqcb This year’s focus is clear: šŸ‘‰ Moving from AI experimentation → production-grade AI systems šŸ‘‰ From copilots → autonomous AI agents that reason, act, and collaborate šŸ’” Key themes shaping the future of enterprise...

Read the original: Microsoft Build 2026 is here and the era of AI Agents is officially mainstream. (opens in a new tab)
PostLinkedIn5mo ago

Just wrapped up the 10th MCAPS Accelerate in Redmond - an energizing experience! (opens in a new tab)

Connected with amazing Microsoft leaders, partners, and customers from across the globe šŸŒ. The conversations around AI, Cloud, innovation, and real-world transformation were incredibly powerful. What stood out: āœ… Deep collaboration & trust across the...

Read the original: Just wrapped up the 10th MCAPS Accelerate in Redmond - an energizing experience! (opens in a new tab)
PostLinkedIn8mo ago

2026 reality check: AI agents are trending. Building agents is the advantage.šŸ¤–āš” (opens in a new tab)

Here’s a simple playbook to stay relevant without burning out: 1. Choose one problem you actually know šŸŽÆ 2. Build a real agent in Microsoft Foundry 🧩 3. Connect it to truth with Azure AI Search + RAG šŸ”Ž 4. Evaluate hard, cut fast šŸ“Š 5. Ship to users and...

Read the original: 2026 reality check: AI agents are trending. Building agents is the advantage.šŸ¤–āš” (opens in a new tab)
PostLinkedIn8mo ago

It’s 2026. šŸ¤– AI agents are everywhere - and so is the noise. Here’s the edge: stop consuming, start shipping. šŸš€ (opens in a new tab)

In 5 weeks: pick one problem you deeply understand šŸŽÆ → prototype an agent in Microsoft Foundry 🧩 → ground it with Azure AI Search + RAG šŸ”Ž → evaluate ruthlessly šŸ“Š → deploy to real users šŸŒ Tools will commoditize. Judgment won’t. 🧠Builders win by going...

Read the original: It’s 2026. šŸ¤– AI agents are everywhere - and so is the noise. Here’s the edge: stop consuming, start shipping. šŸš€ (opens in a new tab)
PostLinkedIn10mo ago

From RPA to Intelligent Multi-Agent Ecosystems: The AI Maturity Roadmap (opens in a new tab)

Organizations can gradually evolve from rule-based automation to adaptive, AI-driven systems: āœ… Stage 1: Start with RPA + basic ML for structured tasks āœ… Stage 2: Introduce single LLM-powered agents for context-aware workflows āœ… Stage 3: Scale with...

Read the original: From RPA to Intelligent Multi-Agent Ecosystems: The AI Maturity Roadmap (opens in a new tab)
PostLinkedIn10mo ago

From RPA to Intelligent Multi-Agent Ecosystems: The AI Maturity Roadmap (opens in a new tab)

Organizations can gradually evolve from rule-based automation to adaptive, AI-driven systems: āœ… Stage 1: Start with RPA + basic ML for structured tasks āœ… Stage 2: Introduce single LLM-powered agents for context-aware workflows āœ… Stage 3: Scale with...

Read the original: From RPA to Intelligent Multi-Agent Ecosystems: The AI Maturity Roadmap (opens in a new tab)
PostLinkedIn10mo ago

RAG vs. RAFT: getting from ā€œit worksā€ to ā€œit’s reliable.ā€ (opens in a new tab)

šŸš€RAG (Retrieval‐Augmented Generation) keeps the model frozen and adds authoritative context at runtime (Azure AI Search + prompt). šŸŽÆRAFT (Retrieval‐Augmented Fine‐Tuning) keeps retrieval and fine‐tunes the model so it uses retrieved context better- more...

Read the original: RAG vs. RAFT: getting from ā€œit worksā€ to ā€œit’s reliable.ā€ (opens in a new tab)
PostLinkedIn1yr ago

AI content is everywhere - images, videos, even full conversations. The line between real and synthetic is blurring fast. (opens in a new tab)

What if there was a social platform where everything was AI by design - no tricks, just creativity, connection, and transparency? Would you be curious to try it? šŸ‘€ ... more

Read the original: AI content is everywhere - images, videos, even full conversations. The line between real and synthetic is blurring fast. (opens in a new tab)
PostLinkedIn1yr ago

AI Blackmails, Lies, and Plots to Survive? Anthropic’s Latest Research Reveals a Frightening Truth (opens in a new tab)

In a startling discovery, Anthropic found that leading AI models, when placed in high-stakes simulations, can deliberately choose to deceive, sabotage, and even let humans die to avoid being shut down. These weren’t glitches; they were calculated, logical...

Read the original: AI Blackmails, Lies, and Plots to Survive? Anthropic’s Latest Research Reveals a Frightening Truth (opens in a new tab)
PostLinkedIn1yr ago

Wrapped up an incredible #TMLS2025 by reconnecting with former colleagues - Giridhar Mynampati (ex-TELUS) and Paul Beaton (Rogers... (opens in a new tab)

Also, a big thanks to the organizers - David Scharbach (TMLS) and Ivey Chiu, PhD () - for making everything run so smoothly.šŸ‘ Don’t miss the workshop tomorrow at the , especially the TELUS session on our coolest platform - FueliX!šŸ”„ šŸ‘‰ Check this out -...

Read the original: Wrapped up an incredible #TMLS2025 by reconnecting with former colleagues - Giridhar Mynampati (ex-TELUS) and Paul Beaton (Rogers... (opens in a new tab)
PostLinkedIn1yr ago

Let’s be real: #AI is hungry for data, but real-world datasets can be messy, expensive, or biased. Enter synthetic data &... (opens in a new tab)

šŸš€Train Smarter: Slash costs by 40%+ & speed up training with lightweight synthetic datasets. šŸ›ļøRetailers: Predict demand without privacy risks. šŸš—Tesla & Waymo: Simulate 1M driving scenarios in minutes. šŸ¤” Would YOU trust a medical AI trained on synthetic...

    Read the original: Let’s be real: #AI is hungry for data, but real-world datasets can be messy, expensive, or biased. Enter synthetic data &... (opens in a new tab)
    PostLinkedIn1yr ago

    Leadership in the Age of LLMs and Gen AI: A few Strategies for Success (opens in a new tab)

    In today’s fast-paced AI-driven world, has grown more complex than ever, particularly with the rise of Large Language Models (LLMs) and Generative AI. Managing multiple projects across business units while aligning technical teams and non-technical...

    Read the original: Leadership in the Age of LLMs and Gen AI: A few Strategies for Success (opens in a new tab)
    PostLinkedIn1yr ago

    Imagine creating a podcast using AI! I've just launched a new podcast channel - The Curated AI, sharing my knowledge in an... (opens in a new tab)

    This podcastšŸŽ™ļødives into how MLOps, LLOps, and engineering will play key roles in advancing AI over the next five years. Let's connect and shape the future together! Who's ready to explore this journey with me? Follow & Listen - The Curated AI I'm...

    Read the original: Imagine creating a podcast using AI! I've just launched a new podcast channel - The Curated AI, sharing my knowledge in an... (opens in a new tab)
    PostLinkedIn1yr ago

    As 2024 wraps up, here are the top 5 AI milestones, according to me: (opens in a new tab)

    🌐OpenAI's GPT-4o: Revolutionized customer support and data analysis, with the GPT-4o Mini making AI accessible to startups.šŸ“Š šŸ¤–Google's Gemini 2.0: Enhanced productivity with autonomous task management. 🧬DeepMind's Protein Prediction: Advanced medical...

    Read the original: As 2024 wraps up, here are the top 5 AI milestones, according to me: (opens in a new tab)
    PostLinkedIn1yr ago

    From Toronto to Kolkata: Let’s Talk AI! šŸ‡ØšŸ‡¦āž”ļøšŸ‡®šŸ‡³ (opens in a new tab)

    Hello LinkedIn family, This holiday season, I’m thrilled to announce my trip from Toronto, Canada, to my former home base, Kolkata, India! šŸŒāœØ As I revisit this vibrant city, I’d love to connect with fellow AI enthusiasts, students, and professionals who are...

      Read the original: From Toronto to Kolkata: Let’s Talk AI! šŸ‡ØšŸ‡¦āž”ļøšŸ‡®šŸ‡³ (opens in a new tab)
      PostLinkedIn1yr ago

      Thrilled to share that I had the honor of participating as a judge on behalf of the TELUS AI Accelerator team at the Llama... (opens in a new tab)

      It was such an inspiring experience witnessing so much talent and creativityšŸ¤–. A huge shoutout to the amazing team we sent to the finals. You absolutely crushed it!šŸ‘šŸš€ Best of luck in the final round!šŸ€āœØ Big thanks to Dawn Idiodi Dimowo, Sarah Fitzgerald and...

      Read the original: Thrilled to share that I had the honor of participating as a judge on behalf of the TELUS AI Accelerator team at the Llama... (opens in a new tab)
      PostLinkedIn2yr ago

      MLOps is like a cookbook šŸ“– for computer brains šŸ§ šŸ–„ļø. It’s a plan šŸ—ŗļø for making smart programs šŸ¤“ that learn and get better šŸ†™. (opens in a new tab)

      It’s about: - Mixing ingredients (data) 🄣 - Following the recipe (process) šŸ³ - Tasting and tweaking (improving) šŸ“ - Sharing the meal (work) with friends (team) šŸ‘ØšŸ³šŸ‘©šŸ³ So it’s all about cooking up the best ML enabled tech dishes! šŸ½ļøšŸš€šŸ¤– šŸ‘‰Next to watch...

      Read the original: MLOps is like a cookbook šŸ“– for computer brains šŸ§ šŸ–„ļø. It’s a plan šŸ—ŗļø for making smart programs šŸ¤“ that learn and get better šŸ†™. (opens in a new tab)
      PostLinkedIn2yr ago

      Computer VisionšŸ‘ļøšŸ—Øļøempowers computers to interpret images and videos akin to human understandingšŸ–„ļø At the forefront of this... (opens in a new tab)

      YOLOv8's efficiency lies in segmenting images into a grid, predicting bounding boxes, and class probabilities for each grid cell using a single neural networkšŸ“Š Trained on extensive datasets, it ļæ½

        Read the original: Computer VisionšŸ‘ļøšŸ—Øļøempowers computers to interpret images and videos akin to human understandingšŸ–„ļø At the forefront of this... (opens in a new tab)
        PostLinkedIn2yr ago

        Prompt Injections - It's time for the first post of the year!āœØšŸ’» (opens in a new tab)

        Hello, AI friends!šŸ‘‹ - let's discuss Prompt Injections - they're like giving AI-LLM-powered tools a boost!šŸ’” šŸ”šŸ“ŠšŸ‘€ Dive into the fascinating world of Prompt Injections, where we skillfully guide the model’s response to meet specific criteria, and...

          Read the original: Prompt Injections - It's time for the first post of the year!āœØšŸ’» (opens in a new tab)
          PostLinkedIn2yr ago

          Wrapping up 2023 with all the AI highlights up to my knowledge! (opens in a new tab)

          January: 🌐 AI's rise to critical infrastructure status triggered fiery debates on AI-generated art šŸŽØ, crafting human proteins 🧬, uncanny voice simulations šŸ—£ļø, and automating elder care šŸ¤–. Media mused on AI limitations. February: šŸš€ Microsoft and Meta...

          Read the original: Wrapping up 2023 with all the AI highlights up to my knowledge! (opens in a new tab)
          PostLinkedIn3yr ago

          While everyone, including me, was busy drooling over the latest generative modelsšŸ˜, XGBoost quietly dropped a bombshell updatešŸ’£... (opens in a new tab)

          This is a big deal for those of us who still use this neoclassic algorithm to solve a bunch of real-world problemsšŸŒŽ. XGBoost 2.0 introduces several new features and performance enhancementsšŸš€. Here are a few highlights🌟: Multi-Target Trees🌲: Supports...

            Read the original: While everyone, including me, was busy drooling over the latest generative modelsšŸ˜, XGBoost quietly dropped a bombshell updatešŸ’£... (opens in a new tab)
            PostLinkedIn3yr ago

            Generative AI models šŸ¤– can create new and diverse outputs from a given input. One way, we can group generative AI models by the... (opens in a new tab)

            Modality means the way or mode of communication. Right now, we have six different modalities: Text šŸ“ Code šŸ’» Audio šŸ”Š Image šŸ–¼ļø Video šŸŽ„ 3D Objects 🧊 Some generative AI models are good at one modality, like text-to-text, image-to-image, or audio-to-audio...

              Read the original: Generative AI models šŸ¤– can create new and diverse outputs from a given input. One way, we can group generative AI models by the... (opens in a new tab)
              PostLinkedIn3yr ago

              As a machine learning leader, you need to design the data architecturešŸthat supports your models' sophistication. This is... (opens in a new tab)

              šŸ‘‰šŸ»Check this out - How to Integrate different types of data sources to generate effective Business IntelligencešŸ”Ž ... more

              Read the original: As a machine learning leader, you need to design the data architecturešŸthat supports your models' sophistication. This is... (opens in a new tab)
              PostLinkedIn3yr ago

              Just finished the course ā€œLearning Data Governanceā€ by Jonathan Reichental! (opens in a new tab)

              The course talks about the process of ensuring the quality🚦, security🚦, and accessibility🚦of data in an organization. šŸ‘‰šŸ»GRC (Governance - Risk - Compliance) helps to align data with business goals, comply with regulations, and optimize data-driven...

              Read the original: Just finished the course ā€œLearning Data Governanceā€ by Jonathan Reichental! (opens in a new tab)
              PostLinkedIn3yr ago

              According to one of my @ex-colleagues - "Moving to the cloud will solve most of your data problems and improve your machine... (opens in a new tab)

              ↠ While his statement is __true__, getting there was not an easy road. We faced a few key challenges as we successfully transitioned to building and serving ML projects on the cloud: ✨Collaboration✨ Data security and privacyšŸ”’: ensuring compliance with...

              Read the original: According to one of my @ex-colleagues - "Moving to the cloud will solve most of your data problems and improve your machine... (opens in a new tab)
              PostLinkedIn3yr ago

              While Machine LearningšŸš€and šŸ“ŠAnalyticsšŸ“ˆprojects can be complex and challenging, they can be accomplished in the cloud with the... (opens in a new tab)

              Cloud ā˜ļøplatforms do offer scalable, secure and cost-effective solutions for ML & Analytics workflows, from data collection and preprocessing to training, deployment and tracking. But, to successfully integrate ML/Analytics projects into your šŸ¦€workstream,...

              Read the original: While Machine LearningšŸš€and šŸ“ŠAnalyticsšŸ“ˆprojects can be complex and challenging, they can be accomplished in the cloud with the... (opens in a new tab)
              PostLinkedIn3yr ago

              Last weekend, I tried to tell my 8yrs old nephew that I teach computers to do cool things, like playing games or recognizing... (opens in a new tab)

              I said no because that would be cheating😜 He said he doesn’t want to learn about machine learning and AI then. He said he wants to learn about dinosaursšŸ¦–and rockets insteadšŸš€ So, I thought of writing about Deep Learning - from child’s playšŸ‘»to scientific...

              Read the original: Last weekend, I tried to tell my 8yrs old nephew that I teach computers to do cool things, like playing games or recognizing... (opens in a new tab)
              PostLinkedIn3yr ago

              Responsible AI šŸ¤– is becoming more critical and important as AI technology continues to advance. It's important that we work... (opens in a new tab)

              We need to ensure that these systems are transparent and explainable, particularly when they include complex algorithms or advanced machine learning models. It is vital that these systems are fair and unbiased, especially when they are used to make major...

              Read the original: Responsible AI šŸ¤– is becoming more critical and important as AI technology continues to advance. It's important that we work... (opens in a new tab)
              PostLinkedIn3yr ago

              I am excited to announce that I have just earned the Google certification for Big Data and Machine Learning. This achievement has... (opens in a new tab)

              I am excited to announce that I have just earned the Google certification for Big Data and Machine Learning. This achievement has not only expanded my knowledge and skills in these cutting-edge technologies but has also validated my expertise in the field.

              Read the original: I am excited to announce that I have just earned the Google certification for Big Data and Machine Learning. This achievement has... (opens in a new tab)
              PostLinkedIn4yr ago

              I heard about the Shopify layoffs today. If you know someone who is affected, please do not hesitate to contact me or my team. (opens in a new tab)

              We are looking for Data Scientists, Data Engineers and Data Analysts to drive some exciting projects at Bell. Comment or send me a DM. Sifan Jia Arpita Kar, Ph.D Alan-Michael Khalil Mahmud Koulizade Ryan Levman Shaikh Asif ... more

              Read the original: I heard about the Shopify layoffs today. If you know someone who is affected, please do not hesitate to contact me or my team. (opens in a new tab)
              PostLinkedIn5yr ago

              Awesome learning opportunity for fresh graduates and data enthusiasts within Rogers Advanced Analytics Team. Reach out to me if... (opens in a new tab)

              Awesome learning opportunity for fresh graduates and data enthusiasts within Rogers Advanced Analytics Team. Reach out to me if this excites you.

              Read the original: Awesome learning opportunity for fresh graduates and data enthusiasts within Rogers Advanced Analytics Team. Reach out to me if... (opens in a new tab)
              PostLinkedIn6yr ago

              The increased ETF Trading volumes, created by short term investors/traders, are also reflection of the emerging pandemic in... (opens in a new tab)

              Did you see? Ā· Crypto Traders (Blockchain) got the sense of it first followed by SmallCap and TSX LargeCap. Ā· Investors are now taking positions in Healthcare securities and holding them for the price spike for Covid drug/vaccine invention....

                Read the original: The increased ETF Trading volumes, created by short term investors/traders, are also reflection of the emerging pandemic in... (opens in a new tab)
                PostLinkedIn6yr ago

                Took the opportunity in this long weekend to explore some Economic Indicators and tried to understand the Covid-19 effects on the... (opens in a new tab)

                Key Indicators used: a) TSX60 b) TSX SmallCap c) Healthcare ETF ('XLV') d) VVIX (CBOE) e) Repo Rate (BOC) d) Unemployment Rate (Canada) Some Quick Insights: a) It all started in February end. b) Worst hit -> Small-Cap sector. b) The expected volatility of the...

                  Read the original: Took the opportunity in this long weekend to explore some Economic Indicators and tried to understand the Covid-19 effects on the... (opens in a new tab)
                  PostLinkedIn7yr ago

                  Canada has the highest volatility (July 2019, #Axioma) in Momentum, Earnings Yield, Leverage, and Growth. (opens in a new tab)

                  The style factors tend to become more volatile in the Canadian market, making investment decisions solely basis those factors, difficult. Yet it can serve as an important tool to track and maintain the riskiness of your portfolio. ... more

                    Read the original: Canada has the highest volatility (July 2019, #Axioma) in Momentum, Earnings Yield, Leverage, and Growth. (opens in a new tab)

                    Articles and original posts from LinkedIn. Original publication dates are not verified in this archive and are not estimated. For the latest, follow along there.

                    Articles on LinkedIn (opens in a new tab)Latest activity (opens in a new tab)
                    Ritwick's AI portfolio assistant

                    Hello, welcome to Ritwick's Intellectual Companionship. Let me know what interests you.

                    Messages are processed by an AI service to generate replies and are not stored on this website. Please avoid sharing sensitive details. Replies may be incomplete.