Dhanush Ravuri

Flipkart is one of India's largest e-commerce platforms, offering a wide range of products across categories like electronics, fashion, home, and groceries. Founded in 2007 and now a subsidiary of Walmart Inc., Flipkart serves hundreds of millions of customers and is known for innovations like cash-on-delivery, easy returns, and Big Billion Days. It plays a key role in shaping the digital retail experience in India.

• Achieved $350K annual cost savings by implementing a high-throughput Redis powered Image-Similarity-Platform searching over 10M+ image embeddings for top-K matches, attaining p99 within 200ms.
• Improved accuracy of Image-Similarity-Platform from 88% to 96%, by transitioning from LSH-based architecture to Qdrant vector database, after benchmarking Milvus and Elasticsearch, reducing p99 latency from 200ms to 75ms at 150 QPS.
• Delivered a 13% improvement in Fraud-Recommendation-Engine service latency at 180 QPS by upgrading from Java 8 to 17 and optimizing JVM performance by G1GC to ZGC migration using JFR and Prometheus.
• Developed a centralized configuration management service using Spring Boot and MySQL for deployment-free configuration updates, exposing CRUD via REST APIs, with RBAC, promotion/rollback, and a Cron for 5 minutes configuration refresh.
• Built a backtesting application with Apache Spark processing 12M+ daily records from HBase in under 10 minutes, enabling rapid performance evaluation of updates to configuration files, with auto-published summary reports to GCS.
• Led the end-to-end data center migration of a 480TB Hadoop cluster across 50+ nodes; Also migrated team-managed HBase to central offering without any downtime, involving 60+ tables, 10+ applications, and over 300TB of data.

Spur.fit is a fitness technology platform that leverages advanced computer vision and AI techniques to deliver personalized workout experiences. It focuses on enabling users to track their fitness goals with innovative solutions like real-time motion analysis, repetition counting, and privacy-enhanced live video processing.

• Improved exercise pose classification model accuracy by 18% through building a custom CNN model on top of body key-points extracted from PoseNet and MoveNet, refining exercise phase detection for accurate repetition counting.
• Integrated real-time human background segmentation in live workout videos using OpenCV, employing MOG2 background subtraction and GrabCut, achieving 24 FPS at 720p and masking 94% of background pixels to enhance user privacy.

The 10x Academy is an intensive coding bootcamp designed to transform individuals into industry-ready software developers. It offers hands-on training in data structures, algorithms, and full-stack development through real-world projects and personalized mentorship.

• Mentored around 200 students in improving their problem solving skills and conceptual clarifications on DSA and Web development.
• Lectured solutions regularly and conducted one-to-one interactive sessions to effectively facilitate their learning and growth.