At VERICODE, I architected VeriHire, an internal recruitment platform powered by LLMs and designed to run on-premise. I structured its delivery roadmap and designed document-processing and natural-language-processing pipelines for resume ingestion and candidate matching.
For VeriHire, I modeled semantic search using Qdrant and FAISS, and developed search and recommendation engines to match queries and job descriptions with candidate resumes.
At VERICODE, I also built monitoring and telemetry infrastructure with Prometheus and Grafana, and trained an XGBoost model to forecast server capacity and resource exhaustion.
On my ongoing openHiRe project, I’m architecting an open-source ATS focused on local inference and infrastructure optimization. I also developed a PDF restoration pipeline that uses OCR and layout detection to produce structured Markdown.

