performance engineering · backend development

Building production-grade systems from research prototypes.

I do my best work when I'm finding ways to extract that last bit of performance from cloud infrastructure, surfacing LLM failures to business analysts, and writing inference code to run at the edge.

in/kworathur kworathur [email protected]

43%

increase in hourly job throughput after prediction engine refactor

16K

protein structures processed in parallel by feature extraction pipelines

8 hours

saved each month by publishing API endpoints for chemical property prediction

Open to new opportunities

Get in touch →

Writing

What a Broken Benchmark Taught Me About Reproducible Experiments May 2026

A story about chasing datacenter energy savings that seemed to good to be true.

Performance
Distributed Tracing
Jaeger
Linux

Experience

Software Engineer Intern | May 2026 – Present

Technologies: AWS, Terraform, Python

Refactored the prediction engine to use SQS, increasing hourly job throughput by 43%.
Automated production backup testing and restoration using Python scripts and Terraform, increasing testing coverage from less than 1% of platform data to >80% and ensuring SOC-2 compliance.

Software Engineer Intern | Sep 2024 – Jul 2025

Technologies: PyTorch, FastAPI, React

Deployed inference endpoints on AWS SageMaker, saving roughly 8 hours each month in coordination overhead.
Built pre-processing pipelines ingesting 16,000+ protein structures for drug discovery research.