AI Resume Screener for Niche Technical Roles

Engineering hiring managers at startups face a specific pain: for niche technical roles (embedded firmware, ML infra, Rust systems, FPGA design), keyword search resume tools flood them with false positives. Someone who lists "Python" and "machine learning" may be nowhere near the actual skill depth required, and someone who describes the same competency in different terms gets filtered out. The community signal here is oblique but real - the HN thread about a rejected candidate obsessing over a decline reflects how opaque and frustrating the screening pipeline is on both sides. SkillGraph (the proposed product) ranks and filters applicants for specialized engineering roles using skill-graph matching rather than keyword search. Instead of counting term overlaps, it maps a candidate's demonstrated skills against a graph of adjacent and prerequisite competencies, so a Rust systems engineer with C++ and memory-safety background surfaces even if the exact job keywords are absent. The solution: ingest job descriptions and resumes, build a skill graph per role, and produce a ranked shortlist with explainable match reasoning. MVP is narrow - support 3 to 5 role archetypes (embedded, ML infra, backend distributed systems), let a hiring manager paste a JD and upload a batch of resumes, and return a ranked list with per-candidate skill-gap notes. Revenue is a per-seat or per-role SaaS subscription aimed at startup hiring managers who lack a full recruiting team. GTM is founder-led outreach into startup engineering leaders via YC and technical hiring communities, plus content on why keyword ATS screening fails for specialist roles. Growth loops come from hiring managers sharing shortlists with hiring committees (exposing more managers to the tool) and from candidates who get better-matched invitations advocating for the product. Honest caveat: the keyword vertical shows steep decline and the market is crowded, so the wedge must be defensibly narrow.