Headline

Data Scientist @WebMD| IIT Patna | Building LLM Applications, RAG Systems & Multi-Agent AI Solutions

About

AI Engineer with experience building LLM-powered applications, RAG systems, AI agents, and healthcare AI solutions. Expertise: • Python, FastAPI, Django • LLMs, LangChain, LangGraph • Vector Databases, ChromaDB • Prompt Engineering and Evaluation • SQL, BigQuery, Cloud Technologies Currently building AI systems that transform natural language into production workflows and analytics solutions.

Location

Bengaluru

AI stack

claude code

codex

cursor

langgraph

opencode

claudecode

Work experience

Data Scientist at WebMD, Dec 2024 - Present · 1 yr 9 mos, Remote, Architected scalable analytics pipelines using Python, SQL, Google BigQuery, and Vertica to process large-scale pharmaceutical datasets (Rx, Claims, DX), reducing campaign measurement turnaround time by 60% (from 5 to 2 days). Developed a GenAI-powered analytics assistant utilizing LangChain and LLMs to automate text-to-SQL translation for business queries, achieving 85%+ accuracy and reducing manual ad-hoc workload by nearly 60%. Engineered geospatial clustering algorithms to construct "geodomes" from raw HCP address data, optimizing field force deployment by identifying high-density physician clusters and reducing travel inefficiencies. Led statistical impact measurement for multi-channel digital campaigns using ANCOVA modeling and test-control matching to quantify incremental NRx/NBRx lift and ROI for pharmaceutical clients.

Trainee Data Scientist at Rakuten India, May 2024 - Dec 2024 · 8 mos, Bengaluru, Developed a specialized Financial BERT model for short-text classification, achieving 91.35% accuracy and significantly outperforming the baseline JLBERT model (88.05%). Fine-tuned Japanese LLMs (Mistral, ELYZA-Llama-2, Youri) for domain-specific financial tasks, optimizing model performance to reach 91.05% accuracy. Engineered a Multi-Agent System using CrewAI, LangChain, and OpenAI API for the "AI Powered E-Commerce Consultant," which was shortlisted in the Top 100 at Rakathon. Built and deployed the Genre Classification Portal, a production tool enabling bulk merchant categorization with confidence scoring and active feedback loops.

Postgraduate Researcher at AI-NLP-ML Research Lab IIT Patna, Jul 2023 - May 2024 · 11 mos, Patna, Developed NLP classifiers using BERT-Base and DistilBERT to detect and categorize patient complaints with 82.95% accuracy. Leveraged GenAI for Healthcare: Implemented and evaluated BioMistral, Falcon, and Zephyr models to handle complex linguistic nuances in medical feedback. Data Engineering: Built a robust scraping framework with Python and BeautifulSoup to aggregate 7.5k raw reviews for model training. Managed Data Annotation: Oversaw the creation of a high-quality "Gold Standard" dataset through rigorous human annotation protocols.

Training and Placement Co-ordinator at Indian Institute of Technology, Patna, Jun 2023 - Apr 2024 · 11 mos, Patna, Bihar, India

Data Analyst at Thermax Limited, Dec 2019 - Dec 2020 · 1 y, Pune, Maharashtra, India

Education

Master of Technology - MTech, Artificial Intelligence, Indian Institute of Technology, Patna

Bachelor of Engineering - BE, Maharashtra Institute of Technology

Public links

github: https://github.com/pratik1208