Back to Projects
2024LLM Guardrail Gateway
Enterprise AI Resume Checker
Hardened full-stack microservice with LLM guardrails for semantic resume parsing.
Technologies Used
PythonFlaskGroqLlama 3.3PydanticVercel
Built a hardened full-stack microservice that semantically maps candidate qualifications against job description requirements using the Groq LPU inference engine running Llama 3.3 70B.
Designed a three-layer LLM Guardrail Gateway: a regex-based prompt injection scanner, a PII redactor under GDPR principles, and a Pydantic BaseModel schema validator for deterministic JSON output.
Deployed as a Vercel Serverless Function with automatic scaling and minimal cold start latency, serving both static frontend assets and the Python worker from a single vercel.json configuration.