talent-search — Verified AI Talent Sourcing

Manual talent sourcing for high-stakes technical roles often results in noisy lists, unverified identities, and halluncinated credentials. This skill provides a disciplined, multi-channel search framework for technical recruiters and engineering managers targeting AI, infrastructure, and quantitative finance talent. It replaces vague keyword searches with a rigorous verification process that demands primary source evidence for every candidate claim.

What it does

• Cross-channel sourcing executes parallel searches across academia, GitHub, target company tech blogs, and professional communities.
• Identity verification enforces a two-source rule to link GitHub accounts, paper authorships, and corporate roles to specific individuals.
• Library management maintains a local JSONL database of candidates to enable incremental updates and prevent duplicate outreach.
• Signal translation converts job descriptions into "strong signals" like core repo contributions or first-author papers rather than just resume keywords.
• Tiered reporting produces Excel deliverables categorizing candidates by confidence level and adherence to school or location bars.

How it works

1. Parameter Lock: Define the persona, graduation window, school bar, and location constraints based on a JD or benchmark profile.
2. Channel Dispatch: Search targeted channels using sub-agents to gather leads from repositories, arXiv, and industry rosters.
3. Verification: Perform single-pass or five-pass cross-verification to confirm identity and education against primary sources.
4. Export: Run gen_excel.mjs to generate a formatted candidate list with verified highlights and contact clues.

Frameworks & tools

The tool uses Node.js (v18+) for data processing and Excel generation. It integrates with web search APIs, GitHub, and browser automation tools for login-walled sites.

Why this beats prompting it yourself

Generic prompts often return outdated profiles or hallucinate career histories. This skill implements a strict verification logic that discards any candidate who cannot be bi-directionally verified through primary sources like paper authorship or commit logs.

Use cases

• Mapping the core contributors of a specific open-source LLM framework.
• Building a pipeline of PhD graduates from C9 or Ivy League labs with specific industry tenure.
• Verifying a list of names from a competition roster against their current professional roles.
• Monitoring target company GitHub organizations for new high-signal talent arrivals.

Known limitations

Does not support real-time access to private LinkedIn data without local browser automation. Graduation years for industry engineers are often inferred rather than stated facts.