Topic (workflows)
I use GenAI where it creates leverage: generating structured artifacts, accelerating iteration loops, and reducing repetitive engineering work—while keeping an evidence trail so results remain trustworthy.
- Goal: faster iteration without losing rigor
- Approach: tool-centric pipelines · structured artifacts · reproducibility
What I believe works (and what doesn’t)
Works
- Constraining generation with structured schemas
- Grounding with project context (specs, IRs, signal allowlists)
- Using LLMs for draft → critique → repair loops with tooling in the loop
Doesn’t
- “Prompt-only” pipelines without artifacts or traceability
- Generating assertions/tests without environment constraints
- Treating LLM output as truth without validation
Where this shows up on my site
- SaxoFlow case study (tooling + workflow design)
- Projects (practical GenAI engineering work)
Evaluating GenAI for verification or EDA tooling?
If you want to collaborate, discuss roles, or review artifacts, reach out.