Hi, I’m Shamil Sherhan.
I’m an AI/ML engineer who loves building things — from multi-agent systems to craft toys, and whatever else catches my curiosity. I write about all of it here, from real-world engineering lessons to weekend side quests.
Available for
AI/ML roles · consulting · collaborations
Recent Writing
Testing Non-Deterministic Agents in CI
A practical design for combining deterministic contracts, semantic checks, DeepEval, Promptfoo, and dual LLM judges without letting flaky evaluation hold every pull request hostage.
Case Studies
Building an Evidence-First Knowledge Assistant
How a basic internal search interface evolved into a conversational assistant with source authority, evidence-level citations, durable history, and bounded live retrieval.
Generating UI Prototypes from a Living Design System
A UI-first workflow that snapshots the real application, converts screens into compact semantic templates, and lets an LLM make bounded changes instead of inventing the product from scratch.
Elsewhere
Personal notes, objects, media, places, and small experiments.