- Managed and mentored a team of two software engineers, overseeing task assignments, roadmap planning, and professional development to drive project success and team growth. - Developed an innovative solution for updating DBC parsers, enabling automatic detection of updates between versions of DBC files with optional user inputs. This tackled a complex industry challenge where direct detection of differences isn't straightforward. This reduced DBC Parser update times from few days to few minutes. It freed up 30% of bandwidth of engineering team. Upgraded the DBC parser architecture by transitioning from single-process deployments to Kubernetes pods. This shift enhanced resource utilization and ultimately cost reduction. - Scaled the CAN parsing algorithm to handle real-time data from over 1,000 devices emitting 1,200 data points per second. Leveraged object storage solutions like Amazon S3 and Azure Blob Storage to design a custom message queue system, overcoming bottlenecks associated with traditional databases. - Collaborated closely with clients to gather requirements, translating abstract needs into concrete product features. Authored Product Requirement Documents (PRDs) and worked with UI/UX teams to design intuitive user interfaces, ensuring customer satisfaction and product usability.

- Converted the complex process of parsing Controller Area Network (CAN) data using DBC files into anaccessible feature on the Bytebeam Console. This automation reduced customer onboarding time from one day to instant setup, freeing up 80% of the integration team’s bandwidth.- Incorporated binary search techniques into the CAN parsing algorithm, reducing computational complexity from O(nm) to O(nlog(m)), which dramatically increased throughput.- Developed a chain reduction algorithm for session tracking, reducing input data size by an average of 65\%. This optimization improved data pipeline throughput.- Led a team of two software engineers, promoting best practices like test-driven development and writing clean, idiomatic code, which improved code quality and team productivity.

- Created algorithms to track sessions on IoT time-series data and parse CAN data in real-time. These algorithms accounted for out-of-order data and supported scaling through sharding, enabling efficient real-time analytics.- Leveraged ClickHouse, a high-performance columnar database, for efficient storage and retrieval of time-series data. Effectively used ClickHouse table engines to optimize for analytical queries and simulate message queue functionalities.- Worked closely with clients to refine algorithms based on their input, ensuring the solutions met real-world requirements and could be seamlessly integrated into their data pipelines.

- Instrumented Prometheus metrics for existing Flask services, capturing critical performance metrics to monitor system health effectively.- Built comprehensive dashboards in Grafana to visualize system performance. Also set up alerts for system critical metrics.- Became familiar with tools and technologies such as Presto/Trino for distributed SQL querying, GitHub Actions for CI/CD automation, Querybook for collaborative querying, and Flask for web development

Teaching Assistant for the following courses:1) IC252 Data Science Lab Sessions.2) CS310 Introduction to Communicating Distributed Processes (OS and Networking course)3) CS302 Paradigms of Programming

Explored transfer learning methods on air pollution forecasting using LSTM models to improve prediction in areas with little data. Tasks involved identifying clusters of data on map and training models on small sized clusters using transfer learning from models trained on nearby larger clusters. Tech: Python, Keras.