Talent infrastructure · sourcing · placement
Talent
Talent infrastructure is the part of fieldbuilding that compounds. The same systems that surface 80,000+ ML researchers and engineers from public signals also tell us which sub-fields are growing, which organisations are losing people, and where the next 50 AI safety hires will come from.
For four years I've built that infrastructure and used it to place candidates into the labs and grantee organisations doing the highest-stakes work — UK AISI, FAR.AI, Apollo Research, GovAI, Anthropic, and others. The work is recruiting on the surface; underneath it's a continuous, instrumented read on the global AI talent market.
SteadRise · Head of Talent & Technical Recruiting · Co-CEO
Headline metrics since March 2024
I lead SteadRise's talent function and have been Co-CEO since June 2025.
Raised for the flagship career-transition program
Hires placed via full-cycle recruiting (incl. 5 funded PhDs)
Offers closed across 6 partners · 12 accepted placements
Days to offer (–27% cost per hire)
Distributed team with real-time funnel analytics
Partner labs, startups, and academic institutions advised
Talent sourcing & intelligence
The most useful candidates aren't searching for jobs
In AI safety the people I want to reach are deep in research labs, open-source projects, and academic groups. I have to go find them, evaluate them on their actual work, and earn their attention. The systems below do that at scale.
ML Research Talent Map
50,000-profile map built from five years of ICLR / ICML / CVPR proceedings, with 2-stage candidate matching using vector embeddings and a weighted talent index.
100+ data sources, 80K-record dataset
Python scrapers ingest conferences, Olympiads, GitHub, Codeforces, and the 3B1B Virtual Career Fair — producing an 80K-record dataset at >90% field-level accuracy.
AI evaluation pipeline (Claude API + Airtable)
Scores candidates across 20+ signals with tri-verdict logic and tier-based shortlisting. Ingests CVs, LinkedIn, and GitHub profiles for grounded verdicts across 15+ role-specific configurations.
Scalable recruiting infrastructure
Cloud SQL + Cloud Run with Ashby ATS and LinkedIn Recruiter RSC integrations. Currently piloting LLM-assisted candidate triage via Anthropic's Model Context Protocol.
Personalised candidate experience
250 candidate papers reviewed quarterly. Employer-brand work via in-person outreach at 15+ AI / ML venues including NeurIPS and ICML.
Olympiad-tier sourcing in under-reached regions
Sponsored IOAA, ICPC, IPhO, IOI, and the Polish Mathematical Olympiad to source elite talent from Eastern Europe, SEA, India, and Australasia.
Inside SteadRise
Global AI Safety Fellowship recruiting funnel
Built and ran GAISF's end-to-end recruiting funnel — deliberately designed to surface non-EA-connected technical and governance talent that AGISF-style fellowships under-reach.
For the fellowship-side write-up — cohort design, outcomes, and TIA-first methodology — see Field Building. This page covers the infrastructure side.
Warm candidates across 5 continents
Partner labs (Anthropic, UK AISI, FAR.AI, +)
Talent surfaced by design
Earlier · before Head of Talent
VP, Talent & Program Director · Sept 2023–Feb 2024
Talent Identification Advisor model
Pioneered an 11-regional-expert + 20-junior-collaborator network embedded in elite tech communities globally — generating 12% of high-signal applications without paid advertising.
Geo-targeted employer-brand campaigns
Ran campaigns at NeurIPS and top-tier universities: 2.3× lift in qualified pipeline at 30% lower CPA than campus career boards.
India campus map across 80 institutions
Mapped India's top 80 tech campuses across 13 states via structured surveys. Built placement channels at IITs, CMI, ISI, and IISc to tap CS toppers and specialised STEM researchers typically recruited by quant firms.
Founder
India AI Safety Initiative
Sep 2022–Aug 2023 · acquired by SteadRise
$65K EAIF pilot grant
- Founded India's first Alignment Research Fellowship — 600+ applicants across 40 STEM universities; 24 selected (top 4%); 10 papers now published or under review.
- Led a 10-person campus recruiting pod for aspiring researchers in interpretability, robustness, scalable oversight, and model evaluations.
- Evaluated a 1,000-member pool of top-decile CS / Physics / Math students on algorithms, ML theory, and research interest. 5 are currently pursuing research independently or at labs like the Mila-Quebec AI Institute.
Featured talent projects
The systems, in detail
AI Candidate Screening — signal-based evaluation at scale→
A two-stage Claude-powered screening pipeline that scores enriched profiles across 20+ signals with citations back to source material.
Weighted Talent Index & JD Fingerprinting→
Analyses 106 job descriptions across 10 skill dimensions to standardise what 'great' looks like per role.
Talent Graph Engine→
A graph platform that connects 100K+ profiles across 500+ public sources.
Nexus — emerging talent from layoff data→
Monitors Reddit, GitHub, and HackerNews to surface experienced engineers during workforce transitions.
Partners
Who I've placed for & advised
UK AISI
FAR.AI
Apollo Research
World Bank
J-PAL
Schmidt SciencesHow this connects to the rest of the work
Talent infrastructure that lives only inside a recruiting agency is a thinner version of itself.
The same talent systems power the placement side of Field Building (AISCF, GAISF, and the Alignment Research Fellowship alumni pipeline), they inform the funding theses I share with grantmakers under Grantmaking & Writing, and they underpin the candidate-facing experiments in Special Projects such as Measuremint, the EA Opportunity Board, Nexus, and the Budhimaan Baccha → RLHF pivot.