Dhanush Ravuri

• Engineered a full-stack MERN based polling application, enabling users to create, vote, and engage in real time, with secure authentication and automated deployments on AWS, serving 1000+ users. • Delivered real-time engagement at scale, enabling live voting, comments, personalized recommendations, geospatial and temporal vote visualizations, and analytics dashboards, resulting in a 22% increase in user session duration. • Architected AI infrastructure, automating 60% of poll generation using LLM pipelines, and building a recommendation system with Deep Learning, Semantic Search, Collaborative and Content-Based Filtering, delivering results in 160ms.

• Developed a parallel file system leveraging gRPC and Protobuf, enabling access across 8+ nodes with file striping for efficient data distribution and supporting file sizes up to 2ˆ28 bytes (∼ 256 MB) under high-concurrency conditions. • Optimized system performance to support up to 2ˆ10 (1024) concurrent clients and manage up to 2ˆ16 (65,536) concurrently open files, by applying token-based synchronization, client-side LRU caching, and multithreaded processing with Pthreads.

• Designed a fault tolerant, Multi-Paxos consensus library in Go, implementing concurrent Proposer-Acceptor coordination and leader election, achieving 120ms median latency per instance on a 5-node cluster. • Ensured high fault tolerance and memory efficiency, correctly reaching consensus with up to 2 node failures, supporting dozens of concurrent instances, and safely garbage-collecting forgotten instances to reduce memory usage.