Chinmaya Samal

• Built a Private Data Q&A Platform to streamline reinsurance submission analysis using RAG-based architecture.
• Parsed complex PDFs, DOCX, Excel files, and internal databases via custom pipelines to extract underwriting insights and reduce review time by 80%.
• Developed a hybrid vector search + agent routing system using LangChain, LlamaIndex, and LangGraph, enabling precise answers from fragmented document sources.

Tools: LangChain, LlamaIndex, LangGraph, Qdrant, OpenAI, FastAPI, AWS

• Developed a Clinical Note Summarization System to automate post-appointment documentation for healthcare providers.
• Used Whisper for speech transcription, OpenAI for summarization, and LangChain to generate structured SOAP notes.
• Reduced documentation time by 70%, improved note consistency, and ensured HIPAA compliance for EMR workflows.
• Delivered scalable APIs using FastAPI and deployed end-to-end on AWS.

Tools: OpenAI, Whisper, LangChain, Python, FastAPI, AWS

• Built a scalable AI platform to analyze 1M+ social posts/day across Twitter, Reddit, and financial news—linking sentiment to stocks and crypto tokens.
• Used OpenAI for sentiment classification, Neo4j for entity graph mapping, and Qdrant for RAG-based insight retrieval.
• Enabled graph-aware natural language queries, reducing research time by 80% and improving signal interpretability.
• Designed a backtesting pipeline to validate sentiment signals against historical price trends, improving prediction accuracy.
• Delivered a production-grade backend using FastAPI, Docker, and PostgreSQL, built to process terabytes of financial data in real time.

Tools: OpenAI, Neo4j, Qdrant, FastAPI, Docker, PostgreSQL

• Core contributor to backend infrastructure supporting appointment workflows, clinician tools, and data pipelines.
• Developed ML models for client–clinician matching, appointment optimization, and billing/claims assistance, leveraging BERT and Transformer-based architectures.
• Built robust data pipelines to process and analyze large volumes of text, audio, and video data, ensuring reliable model performance across diverse formats.
• Helped deliver the MVP and production rollout, integrating AI features into clinician workflows and admin automation.

Tools: Python, FastAPI, PostgreSQL, AWS, BERT, Transformers, Docker

• Founded and led an AI-driven analytics platform to improve public transportation in India using real-time insights and predictive modeling.
• Designed and built data pipelines to ingest and process GPS, fare, and schedule data; applied clustering and forecasting techniques to optimize route efficiency and passenger load distribution.
• Delivered the MVP with a small team, conducted pilot tests with regional transit operators, and iterated based on field feedback.
• Although shut down due to COVID and product-market fit challenges, the venture strengthened my product strategy, technical leadership, and applied ML execution.

• Led data engineering efforts for NSF-funded Smart Cities research, focusing on large-scale transportation analytics and AI-driven mobility solutions.
• Designed and implemented robust data pipelines to process multi-terabyte transportation datasets from public transit, rail, and traffic sensor systems.
• Contributed to models for traffic speed estimation, congestion response to incentives, and transit on-time performance optimization.
• Developed predictive systems for freight train arrival time estimation across U.S. rail networks using spatiotemporal analysis and machine learning.
• Published multiple peer-reviewed papers in top-tier transportation and urban analytics journals.

Tools: Python, SQL, ArcGIS, PyTorch, Tensorflow, PostgreSQL, Pandas, Scikit-learn