Rajat Komawar

I joined JPMorgan Chase & Co. in July 2026.

I developed internal tools for the engineers who plan and build the AWS network. The largest was a repository for more than 450 network device specifications, served through an API with sub-120 ms read latency and paired with a TypeScript interface whose LLM-powered natural-language input resulted in approximately 70% faster schema creation.

I also re-architected the hardware requirements engine as a standalone service that determines hardware for more than 250 datacenter connections, saving 100–200 planning hours per cycle and preventing over $100K in cost overruns. In addition, I reduced the runtime of an ML-based device ordering job from 120 minutes to 4 minutes while expanding its scope from four datacenters to an entire region, and built the DynamoDB Streams, Lambda, and Redshift pipeline that supports analytics across 37 endpoints.

Technologies: Java, Python, TypeScript, AWS (DynamoDB, Lambda, Redshift), and LLM integration.

I contributed to the AWS network cost modeling system by consolidating its backend behind a unified API, which delivered 25% lower latency, and by building a Vue.js interface for submitting, searching, and downloading cost models. I also introduced an automated input-validation framework that produced 15% higher submission accuracy, with 90% test coverage.

Technologies: Python, Flask, Vue.js, AWS S3, and Pandas.

I led a team of three interns in migrating more than 200 workflows from Oozie to a high-availability Apache Airflow scheduler, improving monitoring, reliability, and ease of deployment. We integrated the new scheduler with Cisco's existing Hadoop and Spark infrastructure.

Technologies: Apache Airflow, Hadoop, Apache Spark, Python, and Jenkins.