At AquilaX.ai, I led development of SecuriTron AI, a GPT-2 and LLaMA-based generative AI engine integrated with eight SAST scanners. It delivered structured JSON reports with 96% accuracy and 20% fewer false positives.
I fine-tuned custom LLMs on proprietary code and security datasets, and architected FastAPI model serving with CI/CD pipelines using GitHub Actions and Docker Compose. I also integrated MLflow and Weights & Biases for experiment tracking and model retraining workflows.
During my internship program at Code Clause, I developed and trained a PyTorch CNN for crop disease classification, achieving 92% validation accuracy across 15 disease classes. I’ve also developed an LLM-powered PDF Q&A system and fine-tuned a Google flan T5 model for English-to-Tamil translation.
