CV
Link to PDF resume
Work experience
- Senior Software Engineer, Aptiv, Banglaore, India Jul.2024–Present
- Authored an edge-device communication framework using Google’s Agent2Agent protocol, enabling seamless, low-setup interaction between edge agents over the network for distributed AI applications with minimal latency of 2s end-end.
- Developed a neural network intent classifier achieving 90% accuracy using online learning, clustering, and knowledge distillation techniques (~5% performance drift) to extract actionable insights from chatbot interactions while quantifying system effectiveness.
- Orchestrated a requirement analyzer tool that chunks and embeds incoming requirements into a vector database, which improved hardware system design quality and saved 35% design and review time.
- Created a test case optimization framework using a graph database to map historical test cases, runs, and system requirements—automating test selection by traversing the graph to identify high-risk areas in new requirements saving 100k dollars per year.
- Designed and implemented efficient data pipelines for model evaluation and KPI computation, enabling robust tracking of usage metrics and cost savings across AI systems.
- Software Developer, Oak AI, Boston, MA Jan. 2024 – July 2024
- Conceptualized and spearheaded development of a Retrieval Augmented Generation (RAG) framework enabling a chatbot to deliver contextually relevant responses from documents. Leveraging efficient data storage and indexing with vector database and semantic document chunking.
- Achieved a 20% improvement in database storage efficiency through the implementation of an advanced algorithm and optimized database architecture. This initiative successfully identified and managed duplicate documents and chunks based on text content, enhancing overall data organization.
- Research Intern, Lab for Learning and Planning in Robotics(NEU), Boston, MA May. 2023 – Present
- Multi-Agent Reinforcement Learning: Developed continuous macro-action environments for Multi-Agent Reinforcement Learning. Analyzed and tested continuous macro-action PPO algorithm for robotics and autonomous systems.
- Hierarchical Reinforcement Learning: Achieved state-of-the-art performance by training agents using option-critic and actor-critic algorithms for centralized multi-agent reinforcement learning on four rooms and petting zoo MPE environments.
- Data Engineer, Redbus Jan. 2020 - Jul. 2021
- Big Data Pipelines: Improved customer retention by 10% through the design and implementation of big data pipelines, ETL and parallel processing models for customer life cycle management with cross-functional communication with marketing teams for analysis; leveraging AWS cloud and database technologies (MongoDB, Cassandra, Postgres).
- Distributed Parallel Processing in Spark: Slashed processing times by 200% for weekly and monthly statistical analyses implementing Spark programs on AWS platform, enhancing efficiency.
- Flask and Golang Web APIs: Developed low latency Flask and Golang API’s handling millions of hits per second. Optimized legacy api’s to cut down response time by 50%.
- Machine Learning Intern, Allgo Embedded Systems Jun. 2019 - Aug. 2019
- Led a research-based project to identify activities of vehicle drivers given video input from dashboard cam and classify activities into distinct classes in real time using computer vision and deep learning.
- Analyzed and compared multiple models including 2 stream CNNs, LSTM and traditional CV techniques. Demonstrated final model with ability to correctly identify a smaller subset of actions with high accuracy up to 86%.
Education
- M.S in Computer Science , Northeastern University, 2023
- Bachelor of Engineering in Information Science, JSS Science and Technology University (SJCE), 2020
Skills
- Languages
- Python
- Java
- Go
- Scala
- Typescript
- Tools and Technologies
- GenAI
- RAG
- MCP
- AWS
- MongoDB
- Hadoop
- Apache Spark
- Redis
- Docker
- Libraries and Frameworks
- Langchain
- Langgraph
- A2A
- Neo4j
- PyTorch
- Tensorflow
- Django
- Flask
- Pandas