Global Head of Talent Acquisition at kAIgentic, the enterprise intelligence layer. I head global TA and build the talent system for a platform that turns tacit work knowledge into operational intelligence. 20+ years building and scaling TA functions across APAC, North America, and Europe.
Before kAIgentic, I headed R&D hiring at Atlan across Engineering, Product, Design, IT, and Security. What makes me different: I also build the systems underneath hiring. An AI sourcing mapper that turns a job title into a full talent landscape in 10 seconds. Review agents that pressure-test hiring decisions and document structure. A talent evaluation engine for interview decisions. A prompt governance engine for production LLM systems. All shipped, all built with Claude as my engineering partner, and public where it is safe to be. I wrote Raising the Bar, a playbook on building teams that compound strength through every hire.
Tools I ship to make judgment, evaluation, and operating discipline repeatable. All built with AI as my engineering partner. Public where it is safe to be.
|
An agent modelled on Patty McCord, former Netflix Chief Talent Officer. It pressure-tests hiring, pay, performance and exit decisions, and quotes her only from a verified bank.
|
An editor built on Barbara Minto's Pyramid Principle. It tests the structure of any document or deck, then rebuilds it: governing thought, MECE grouping, vertical logic.
|
|
AI talent landscape mapper for recruiters. Type a role, get target companies, adjacent pools, wildcard bets, and ranked titles with live connection lines. Replaces two hours of desk research with one form. Live app.
|
My operating system: master prompt, voice rules, working rules, and review agents. It includes Gabbar, the interview evaluation framework: a hiring doctrine, proof of real work plus coachability, an interviewer playbook and a scorecard. Private.
|
|
The code module of RishiOS: a talent evaluation engine as an MCP server with a 6-dimension rubric, a strict verdict engine (Maybe = No), scorecard validation with auto-repair, a Yes rule for interview decisions, and a culture interview guide. Private.
|
|
The control plane for AI prompts. Score, enforce policy, lock config, audit and route prompts across providers before they reach production. Deterministic, with zero LLM calls inside. Product site.
|
Blind benchmark for LLM reasoning failures. Side-by-side model comparison with hidden identities, factual accuracy scoring, a 10-type failure taxonomy, stability testing, and confidence calibration. 100+ models, 7,500+ responses. Live demo.
|
|
Floating macOS widget that tracks Claude usage and rate limits in real time. Single-file Swift, no dependencies. Site.
|
"Every hire raises or lowers the bar. There is no neutral." — from Raising the Bar