At Ascendion (Nitor Infotech Pvt Ltd), I develop and deploy AI/ML applications across healthcare, finance, and business domains. I built a multi-agent healthcare referral automation system that processes 1,000+ referrals per month and reduced manual processing time by 95%.
I developed a patient readmission risk model using an XGBoost and LightGBM ensemble, achieving 92% AUC-ROC on patient records. I also built a medical document intelligence pipeline that classified documents with 94% accuracy and extracted clinical entities with an 89% F1-score.
My work includes production RAG systems, LLM fine-tuning, and MLOps on GCP and AWS. I’ve reduced LLM API costs by 30% through prompt caching, token optimization, and model distillation, and improved inference latency through runtime conversion and quantization.
