San Francisco, California, United States
• Built an online judge using React.js, Node.js, and TypeScript, with a secure sandboxed environment to prevent execution risks, processing 1M+ submissions and hosting contests for 100K+ users to date, flawlessly under real-time competitive loads.
• Developed a cost-efficient offline processing system using AWS Managed Kafka to handle traffic spikes, a classic case of guerrilla capacity planning, saving $24,192/month. Debugged production issues, improving system reliability during peak load.
• Implemented on-prem system architecture with Nginx and Gunicorn for horizontal scaling, load balancing, and optimized TCP/IP handling, ensuring stable performance during competitive programming contests with 100K+ active users.
• Designed and managed both on-prem and cloud Kubernetes clusters for Code Genie (AI teaching assistant), orchestrating Dockerized workloads across AWS ECS, with integrated RAG search; ensured high performance for 20,000+ weekly learners.
• Enhanced system performance and scalability by implementing Redis for caching, S3 for code submissions, and PostgreSQL for data management, enabling real-time leaderboards and efficient analysis by the Code Genie Al.
• Secured systems using AWS CloudWatch, mitigating DDoS attacks by blocking malicious accounts, achieving >90% uptime, contributing to securing $1M in funding through our pitch to Y Combinator and other investors.
• Extensively administered Linux-based production systems, including kernel-level debugging, process monitoring, and automation via shell scripting for distributed services across on-prem and cloud environments.
• Collaborated on infrastructure migration projects involving virtualized environments (VMware/EC2) and containerized workloads, adapting legacy services into Kubernetes-ready deployments.
• Tech Stack: TypeScript, React.js, Node.js, AWS ECS, Docker, Kubernetes, AWS Lambda, Amazon API Gateway, AWS Managed Kafka, Gunicorn, Nginx, Redis, PostgreSQL, AWS S3, AWS CloudWatch
