From first experiment to production AI.
I work across product strategy, engineering, and applied AI, and I build AI products that hold up in real workflows. Most recently: AI hiring systems for assessments, screening, and conversational interviews.
Systems that reached production
Descriptive titles instead of product names. Built with product and engineering teams; employer-owned.
Enterprise Skills Assessment Platform
Coding and data-science assessments, source-code verification, remote proctoring, and analytics across 500+ skills.
AI-Assisted Candidate Screening
Embedding search and LLM ranking for JD–CV matching, with question generation and structured candidate reports.
Conversational Interviewing System
Agentic multi-turn interviews with dynamic follow-ups, live coding support, and LLM-generated insights for reviewers.
Areas explored across 15+ PoCs
Principles
Workflow fit before model cleverness
Ask whether it makes someone’s day meaningfully easier, not how smart the model is.
Prove it small, then productise
PoC first; carry what holds up into production features.
Platform judgement
Reusable capability vs targeted customisation: making that call well.