AI/ML Engineer specializing in GenAI, RAG Systems, & Healthcare ML. Experienced in designing and deploying production-grade GenAI and machine learning systems. Specialized in LLM-powered applications, semantic search, and agent-based architectures used in real-world environments.
Architected a decoupled LangGraph multilingual routing node with recency-prompt injection for Mediacorp's mewatch OTT chatbot, achieving 100% fallback accuracy on Gemini 2.5 Flash while cutting LLM token costs by 80% compared to upgrading models.
Migrated EPG discovery from NL2SQL to Vertex AI Search, engineering a 15-day rolling BigQuery dataset that eliminated SQL syntax failures and improved search relevance for fuzzy titles and temporal queries.
Designed stateful pagination and context-aware deduplication logic for the VOD recommendation agent, achieving 0% duplicate show suggestions across continuous multi-turn sessions.
AI / ML Engineer, Cloudside Technologies
01/2025 - 04/2026, Coimbatore
Architected an agent-orchestrated financial chatbot enabling natural-language queries over structured financial data.
Designed routing logic across FAQ retrieval, SQL generation, analytics, and forecasting agents.
Designed a modular multi-agent system automating lead generation, enrichment, and multi-channel outreach workflows.
GenAI / ML Engineer Intern, ISPG Technologies
05/2024 - 12/2024, Kochi
Independently architected and deployed a production-grade RAG-based semantic search system for automotive parts discovery.
Built an end-to-end semantic product search engine supporting complex natural language queries with pricing constraints.
Processed 10,000+ product category hierarchies with precomputed BERT embeddings, achieving 100% precision by eliminating false-positive results.
Data Scientist Intern, Microbiological Laboratory Pvt Ltd (MLRS)
01/2024 - 04/2024, Coimbatore
Architected and developed an end-to-end ML system to automate meningoencephalitis pathogen detection from RTPCR data.
Engineered signal-processing pipelines for HRM curves and amplification cycles with feature extraction and ML classification.
Reduced diagnostic interpretation time from hours to minutes, enabling labs to process 100+ samples per day.