AI CV screening and evaluation

• ~45%

  Screening failure rate eliminated by the fixes

  Observed in a 22-pair evaluation run, with fixes covered by 7 regression tests from real error strings. A rerun of that exact evaluation was not recorded.

  Verified

• 0.13%

  Parse failure rate in the later baseline (1 of 800 calls)

  Measured in a later evaluation baseline using a newer model and prompt.

  Verified

• 270+

  Test files added from a zero-test baseline, with evaluation gates blocking deploys

  Based on repository history.

  Verified

• 21.1 s to 12.1 s

  Screening p95 latency with medium rather than high reasoning effort

  Measured in controlled evaluation rounds; the selected setting still passed fairness and injection gates.

  Verified

LLM screening frequently failed and issued hire-or-reject verdicts without evidence. There were no tests or release gates to catch regressions from prompt or model changes.

• 01I traced two production errors to token exhaustion mid-JSON and raw newlines in strings, then addressed them with JSON mode, output caps, a quote-aware extractor and retries.
• 02I introduced the first test suite and CI-gated evaluation harness, using a 40-pair golden set (37.5% in Bahasa Indonesia), counterfactual fairness checks and 20+ prompt-injection classes.
• 03I removed AI hire-or-reject verdicts and added AI disclosure, a system card and identity redaction before model calls.
• 04I adopted passwordless managed identity and HMAC-signed calls between the web tier and backend.

• Next.js
• FastAPI
• Durable Functions
• Azure OpenAI
• Azure AI Foundry
• Document Intelligence
• Cosmos DB
• Key Vault
• Managed Identity
• Entra ID
• GitHub Actions
• Bicep