JustHireMe

JustHireMe is an AGPL-3.0 licensed, local-first desktop app built by Vasudev Siddh. It imports a resume or profile, collects job leads from multiple sources, scores how well each fits, and generates tailored resumes, cover letters and outreach drafts for review. The current release is v1.7.0, with installers for Windows, macOS and Linux.

What It Is

JustHireMe is a job-search workbench that runs on the user's machine. A Tauri desktop shell hosts a React interface and a Python backend sidecar. Profile data, lead history, generated documents, graph data, vectors and settings are stored locally by default.

How the Workflow Runs

A resume (PDF, DOCX, TXT, MD, JSON Resume exports, LinkedIn zips, GitHub or portfolio URLs) is ingested into a Kuzu skill and project graph, with an import report showing what was pulled in, skipped or capped. Source adapters normalize leads from ATS boards, RSS feeds, Hacker News, GitHub, Reddit and custom sources. A deterministic quality gate rejects stale, thin, spammy or senior-only postings. Remaining leads land in a pipeline (discovered, tailoring, approved, applied, discarded) where nothing moves forward without user approval.

Explainable Matching

Ranking combines source signal, lead quality, a deterministic fit rubric, seniority caps, project and stack evidence, feedback learning and semantic similarity from LanceDB vectors. Fit scores are backed by evidence from the user's own graph. Embeddings use a bundled ONNX model (all-MiniLM-L6-v2) with a hashing fallback, so semantic matching works offline without an API key.

Generated Application Materials

For a promising lead, the customizer produces a tailored resume PDF, a cover letter PDF, a founder message, a LinkedIn note, a cold email, a keyword coverage summary and the rationale for selected projects. Drafts are for review, not automatic submission. The README says the approach works for any field and location.

Models, Agents and Setup

LLM steps can run keyless through Ollama, Claude Code CLI or Codex CLI, with optional keyed providers such as OpenAI, Anthropic, Gemini, Groq and DeepSeek. The repository also ships an agent skill and a stdio MCP server exposing score_job_fit, evaluate_lead_quality and extract_lead_intel. Browser automation and auto-apply code is experimental and disabled by default. The installer is thin, and a runtime pack downloads once on first launch. The macOS build is not yet notarized.