Harsha Kanaka Eswar

About Harsha Kanaka Eswar

I’m a Software Engineer currently working at Aira, where I build scalable backend systems and event-driven microservices using Python. I recently developed a high-concurrency ML inference service and automated CI/CD workflows with GitHub Actions, Docker, and AWS.Previously at Samsung R&D, I contributed to Android apps used by over 50M users—redesigning architecture for performance, enabling dual SIM support, and integrating privacy-focused ML features.I hold an MS in Computer Science from UMass Amherst and a BTech in Electrical Engineering from IIT Madras. I\'m passionate about designing efficient, production-grade systems and am currently exploring Backend and SDE roles.Tech Stack: Python, FastAPI, AsyncIO, Docker, Kubernetes, GitHub Actions, ClickHouse, SQL, AWS, Android, Java, Kotlin, JUnit, Mockito

Experience

1. Software Engineer

   Aira Technologies

   Jan 2025 — Present · Saratoga, CA, US

   Built a scalable Python microservice using asyncio, multithreading, and semaphore locks to enable 200+ concurrent ML predictions per minute, significantly reducing latency.• Engineered event-driven backend workflows that improved overall system responsiveness by 60%, ensuring seamless concurrent API execution.• Designed and optimized a high-volume ETL pipeline ingesting over 1 billion records into ClickHouse, reducing query latency by 50%.• Achieved a 70% reduction in data processing time by optimizing batch insert operations, enabling 5x faster ingestion of over 10 million records per day.• Automated CI/CD pipelines using GitHub Actions, integrated SonarQube for continuous code quality checks, and deployed Dockerized applications on AWS, reducing deployment time by 30% and achieving 99.9% uptime.• Orchestrated the deployment of 10+ microservices on Azure AKS using NGINX Ingress controllers and Kubernetes network policies for secure traffic routing; integrated Azure Files and Azure MySQL for persistent, scalable storage.• Developed an interactive chat interface supporting concurrent users, combining WebSockets for real-time updates and SSE streams for LLM-driven reasoning with chain-of-thought display; implemented asynchronous RESTful endpoints with thread-based concurrency and connection pooling, reducing database latency by 40% under load.

Education

• University of Massachusetts Amherst

  Master of Science - MS

  2022 — 2024

• Indian Institute of Technology, Madras

  Bachelor of Technology - BTech

  2016 — 2020