Most job-search tools either leak sensitive profile data to cloud services or flood you with stale, low-context leads. This project treats job discovery as a local, explainable pipeline: ingest leads from many sources, apply a deterministic quality gate, rank fit with profile-aware vectors and rules, then generate reviewable application materials — all stored and executed on the user's machine.
What Sets It Apart
- Local-first, privacy-oriented pipeline — keeps profile graphs, vectors, and generated documents on-device so API keys and personal resumes aren’t sent to third-party services; useful for privacy-sensitive job seekers.
- Deterministic quality gate + explainable ranking — rejects stale/thin/spammy leads and surfaces clear reasons for each score, so users can trust and tune recommendations instead of treating them as black-box suggestions.
- Profile-aware semantic matching — combines a Kuzu profile graph with LanceDB-style embeddings to go beyond keyword matching; this produces role-fit signals informed by projects and skills rather than simple term overlap.
- Reviewable generation (not blind automation) — generates role-specific resume PDFs, cover letters, and outreach drafts for human review; browser auto-apply code exists but is experimental and opt-in.
Who It's For & Trade-offs
Great fit if you want a privacy-first, auditable job discovery flow and are comfortable running a local desktop app with a Python sidecar (or contributing adapters). Ideal for people who value explainability in ranking and want to curate high-signal leads from multiple sources.
Look elsewhere if you need a fully cloud-hosted, zero-setup solution or if you expect turnkey auto-apply automation; packaging and cross-platform installers are active workstreams (Windows is the primary stable installer target). Contributions are encouraged for more source adapters, packaging polish, and enhanced keychain-backed API storage.
