Lore for Recruiting — The portfolio for engineers in the AI era
Resumes don't show AI fluency. Candidates share AI sessions from past projects. No take-home, no extra effort required. Lore scores how they actually think with AI, then matches against your company's target profile.
The problem
Every engineer uses AI. GitHub commits show output, not process. Take-homes show the polished artifact, not how it was made. The real signal lives in the chat threads.
Ten dimensions of agentic engineering fluency
- Context engineering — briefing the model with constraints, files, decisions, and intent.
- Verification and recovery — catching when the model is wrong, understanding why, and recovering well.
- Architectural framing — exploring system-level tradeoffs, not just function-by-function asks.
- Scope control — trimming output to what the problem requires.
- Learning orientation — using AI to go faster on things they understand, not to bypass understanding.
- Delegation judgment — choosing the right supervision level and tool for the task.
- Iteration appropriateness — choosing short iterations for exploration and tighter specs for production work.
- Agent-legible authorship — writing code and docs that future AI and humans can read and modify.
- Eval craftsmanship — building feedback loops that detect when the model is wrong before users do.
- Risk-weighted delegation — knowing when not to reach for AI at all.
How it works
For companies
- Define your company's target profile for the role and stage.
- Candidates upload work they've already done.
- Lore scores each session across the ten dimensions and matches the candidate to your target profile.
- Use sessions as a screener, a technical assessment, or interview material.
For candidates
- Install the Lore CLI so existing Claude Code sessions sync automatically.
- Curate sessions that best show your thinking, with automatic secret scrubbing and optional redactions.
- Apply once and match with companies hiring for your shape across the ten dimensions.
Candidate control and privacy
Candidates choose which threads to share. Secrets are scrubbed automatically, and candidates can redact specific lines or sections. Reviewers see that content was hidden, but not what was hidden.