A point of view
The hard part of AI isn't the model. It's everything around it.
I build the systems that make generative & agentic AI trustworthy in production — the evaluation harnesses, golden datasets, and guardrails that tell you the model is wrong before your customers do. Fourteen years, five companies, tens of millions of users.
What I've learned is all on this site, free. If you're a software engineer trying to move into AI, don't start with a framework — start with the roadmap.
The GenAI Engineer Roadmap
Start here →Moving from software engineering into AI? The material isn't scarce — it's just never in order. This is the order: five stages, built from what I've written doing this work in production, with something to build at the end of each one. Free, and it stays free.
01Foundations02Embeddings & retrieval03RAG in production04Agents & frameworks05Systems & the interviewHow I think about building AI
01 — 0401
Models are commodities. Judgment is not.
The advantage was never the LLM you picked. It's the evaluation harness, the golden datasets, and the guardrails that tell you — reliably, daily — when the system is wrong.
02
Ground it in production reality.
RAG and agents earn trust only when they're measured against real customer utterances — not tidy benchmarks. The eval set is the product spec.
03
Make the black box explainable.
Per-utterance MRR, mAP and Hit-Rate, surfaced so a non-ML stakeholder can debug a bad score themselves. Trust scales when understanding does.
04
Automate the boring parts.
A good agentic workflow closes the vulnerability ticket and opens the PR for review — freeing humans for the ambiguous, high-judgment work only they can do.
Selected work
All case studies →- 01
GenAI Regression Evaluation
A golden dataset grounded in real customer utterances, automated daily runs, explainable retrieval metrics.
Intuit · 2026 - 02
Reinventing the Power Platform
Generative & agentic AI for Power Apps and Power Pages at massive scale.
Microsoft · 2024–25 - 03
Data Protection as a Service
Privacy-by-design microservice on a cloud platform — the work behind my granted patent.
SAP · Patent - 04
Modernizing Enterprise Manager
A 30-year-old product with 10,000+ live customers, brought onto current technologies without breaking anyone.
Oracle · 2022–24
Making the switch
Moving into AI engineering?
The roadmap and every note here are free — start there. When you want your resume torn apart, a real mock interview, or someone keeping you honest month to month, that's mentoring.
See mentoring →Building something
Hiring, or need an architecture review?
RAG evaluation, agentic systems, legacy modernization, or getting a team productive with AI tooling — see consulting, or just tell me what you're building.
Start a conversation →