Jul 2024 – Present · 1 yr 8 mos · Pune, Maharashtra, India

• Developed a real-time transaction decisioning system to prevent fraud, generate alerts, detect velocity breaches, and deliver customer notifications — processing 500K+ requests/day with sub-50ms evaluation latency.
• Optimized transaction processing time by an average of 40% using binary search and efficient XML parsing.
• Accelerated API performance of a Python-based ML service by implementing Redis caching, cutting response time by 75% and timeout errors by 99%. Handled containerization with Docker and deployed to AWS ECS.
• Built a two-way communication system between the fraud detection engine and the mobile app, enabling customers to verify and reattempt risky transactions, increasing approval rates by 30%.
• Implemented parallel Bank-level and Organization-level rule engine API calls using Java Generics, Multithreading and the Template Design Pattern, maintaining low latency and improving system extensibility.
• Built a fraud mitigation API to block/unblock suspicious MIDs at runtime, avoiding frequent redeployments.