At Sysvine Technologies, I improve search and recommendation systems for Elsevier's Pure Portal, building hybrid BM25 and FAISS retrieval across 250k+ research entities and LightGCN personalization for 250k+ interactions. I've also reduced Cassandra calls by 96%, accelerated page loads by 10x, and maintained sub-200ms p99 latency under high-entropy traffic.
Previously at The Hindu Group and Kavi Global, I delivered Solr search upgrades, Sentence-BERT recommendations, FastAPI and Redis services, computer vision moderation, and an AI claims intelligence platform that automated 80% of manual data entry. I enjoy applying practical ML, NLP, and retrieval engineering to measurable product outcomes.
