SAMARTHI LAHIRI

Developed highly parallel data insight generating web platform using RAG, with extensive use of LLMs using relational and non-relational databases and Python FastAPI framework.

• Designed a heuristic algorithm for an NP-hard scheduling and routing problem designed to streamline Medical Sales Rep's visits to doctors with an accuracy of 98.2%.
• Improved API performance of existing codebase by identifying bottlenecks and optimizing code by about 40%.
• Deployed dockerized monoliths in AWS EC2 instance, with a CI/CD Bitbucket pipeline.
• Solely designed, wrote and owned the backend of diff-tracking web solution for Excel sheets in Python.
• Designed modular, maintainable code and wrote exhaustive unit tests.

• Developed a chatbot using ElasticSearch and Python FastAPI.
• Innovated a question suggestion for the chatbot using K-D Trees improving performance by at least 80%.
• Wrote the backend to a web app which lets Pharma affiliate users control versions of their roster and territory spreadsheet files and also marks gaps in the data which reduced manual analyst efforts by about 50%.

• Created an MVP from scratch on the Quorum chain using Solidity, and IPFS.
• Worked on a donation tracking and escrow system that held or rescinded donations based on organizational performance of the receiver.

Ray tracing is an algorithm to mimic the behavior of light rays in the real world to generate ultra-realistic images