Tanurima Halder

Data Scientist with experience developing production-grade Generative AI and LLM-based applications using LangChain and OpenAI GPT frameworks. Engineered RAG pipelines for clinical query resolution, achieving approximately 90% accuracy in grounding responses to official medical guidelines. Proven track record in NLP and deep learning research with multiple publications in peer-reviewed venues like IEEE and ACL.

Expertise in Python-based ML workflows and privacy-preserving machine learning, ensuring data integrity within regulated research environments.

Data Scientist at George Institute for Global Health (2025-07 – Present)

• Developed production multi-turn RAG chatbot using LangChain and OpenAI GPT to assist healthcare workers with complex pregnancy-related clinical queries.
• Implemented retrieval pipelines grounding LLM responses in medical guidelines to ensure high accuracy and regulatory alignment in health queries.
• Built analytics pipelines tracking session duration and query classification accuracy to evaluate model performance and user engagement.
• Engineered reusable prompts and retrieval algorithms for unstructured document ingestion into vector databases to improve response relevance.

Pre-Doctoral Researcher at IIT Bombay (2024-07 – 2025-07)

• Conducted research on privacy-preserving machine learning for medical imaging datasets to maintain data security in regulated environments.
• Developed MRI harmonization models using deep learning to standardize data across disparate medical imaging sources.
• Collaborated with cross-functional research teams in an agile environment to deliver insights and publish findings in peer-reviewed journals.

M.Tech in Artificial Intelligence and Machine Learning – IIT Hyderabad

B.Tech in Computer Science and Engineering – Jalpaiguri Government Engineering College