Graduate Teaching Assistant for Natural Language Processing (CS6120) Spring 2025

As the first engineer at this startup, I ideated, designed and implemented a parenting plan chat-bot using LangChain. Boosted its reliability by leveraging Retrieval Augmented Generation pipeline along with LangGraph Agentic architecture.

Deliverables achieved:

• Fine-tuned LLMs to parenting-specific contexts and built the complete Agentic lifecycle with user in the loop.

• Developed APIs, the backend and frontend, REST APIs using Django & FastAPI complete with React, HTMX based UI.

• Finally, deployed the application on AWS EC2 instance, along with data management services – Postgres DB and Redis.

• Software: Developed Microservices using Spring Boot, Java, REST Kafka messaging to integrate multiple streaming partners (Netflix, DAZN) which helped our clients boost their userbase by onboarding 20,000 new subscribers in the EU.

• Migrated existing codebases from JMS, Tibco to Kafka, Spring and Salesforce APIs easing future developments and integrations.

• Lead the Innovation Team to create utility tools, tailored to the project, that provided a 40% boost to development and test speeds.

• Created a customized Maven archetype for jumpstarting Spring Boot Microservices which eliminated 20% of boilerplate code.

• Standardized dependency management, external API connections and unit tests.

• Provided externalized code vulnerability scanning capabilities by integrating SonarQube.

• Data Engineering: Created customized data pipelines to track sprint progress by processing developer’s GitHub pull request data (comments, code smells, code coverage) using natural language data processing. Built dashboards and automated creation of reports which improved task progress tracking by 60 %.

• Recipient of ACE Award 2022: An accolade for being innovative and going the extra mile.

Designed and developed a Chat-bot integration for automating DevOps tasks with Natural Language. Used Google DialogFlow APIs and Nodejs runtime to understand intents and translate them to Kubernetes commands. Using chat integration, reduced the developer’s time spent on running scripts by 50%. Provided a proof-of-concept for other automation tasks with similar architecture.