At SECURITAI, I develop and maintain Java and Python services that detect PII and sensitive data in unstructured files and databases. I engineered file-format extraction and OCR across PDF, MS Office, OpenDocument, and email formats, and hardened parsers against malformed inputs.
I built and deployed a Python microservice using Gunicorn and multi-stage Docker builds to improve detection-pipeline throughput and scalability. I also re-architected state storage on Redis, hardened Kubernetes workloads, fixed ReDoS flaws, drove CVE remediation, and investigated production OOMs and pod crash-loops.
As a Software Engineer Intern at SECURITAI, I engineered infrastructure for large-scale regular-expression matching using Intel Hyperscan, achieving speedups of 2–3 orders of magnitude. At RETROCAUSAL AI, I developed Django REST Framework API endpoints and automated reporting workflows with storage and retrieval on Amazon S3.
For my final-year project, LLM TEXT DETECTOR, I combined cross-perplexity scores with a fine-tuned T5 transformer to detect AI-generated text, achieving a 94% F1-score on a curated dataset. I earned ICPC Asia Topi Regional gold and silver medals, with national ranks of fourth and eighth, respectively.
