Architecture deep-dives, lessons from production, and field notes from the Microsoft AI stack — Power Platform, Copilot, Dynamics 365, and Azure AI.
-
Learn Fine-Tuning — the hands-on course I’m building during my master’s
Launching a hands-on, self-contained fine-tuning course — 19 lessons covering LoRA, QLoRA, DPO, RLHF, and production. Why I built it, and what surprised me.
-
Authentication patterns for Microsoft Foundry — beyond DefaultAzureCredential
DefaultAzureCredential is the right answer for week one of a Foundry project. By month three you need a real pattern catalogue — managed identities, workload identity federation, OBO, and per-environment role assignment.
-
Starting an Azure Foundry project — the getting-started guide nobody wrote
Five decisions, the day-one auth pattern, a first-sprint checklist, and three traps. The Azure Foundry getting-started guide I wish I’d had before TrafficIQ.
-
Why I built 6 agents instead of 1 mega-agent — lessons from TrafficIQ
Why TrafficIQ ships six specialist agents instead of one 56-tool mega-agent — routing, handoff, MECE tools, and grounding lessons from production.
-
🥈 1st Runner Up — Microsoft Dynamics 365 Customer Insights ITeS Hackathon
Our ITC Infotech team won 1st Runner Up at the Microsoft Dynamics 365 Customer Insights ITeS Hackathon (Feb-Mar 2021) for an intelligent-automation financial-crime solution combining Azure ML, Customer Insights and D365 case management.
-
🏆 Winning Best DEI Use Case at Microsoft HackTogether — Make Life Easy
Make Life Easy was selected as one of four global winners of Microsoft HackTogether: Power Platform Global AI Hack — winning the Best DEI Use Case category.
-
🏆 Winning Best Use of Microsoft Foundry at AI Dev Days Hackathon — TrafficIQ
TrafficIQ won the Best Use of Microsoft Foundry award at the Microsoft AI Dev Days Hackathon — here is the story, the stack, and the 6-agent / 49-tool TRAFI architecture.
-
Building LocalRAG — a fully local AI document search
LocalRAG is a Retrieval-Augmented Generation app that runs entirely on local infrastructure — no cloud LLM, no data egress. Here is how it works and what I learned.
-
Welcome to PowerAI Labs
PowerAI Labs is my engineering notebook in public — architectures, experiments, and the patterns that survive production on the Microsoft AI stack.
PowerAI Labs 