Manish Challa

About

ML/AI engineer with hands-on experience shipping regulated financial AI systems at JPMC and Capgemini, spanning credit risk, fraud detection, and generative AI assistants. Stands out for combining modern LLM/RAG architectures with strong MLOps, real-time infrastructure, and explainability/compliance practices, while delivering measurable business impact in latency, accuracy, cost, and risk reduction.

Key Strengths

• End-to-end ownership of production ML/AI systems from experimentation through deployment and monitoring
• Built generative AI financial assistant with RAG over 1M+ banking records
• Strong credit risk and fraud detection domain expertise
• Improved latency from 5 seconds to under 1.5 seconds through retrieval optimization and caching
• Improved credit prediction accuracy by 25% using imbalance handling and feature engineering
• Reduced manual risk analysis effort by 40%
• Reduced customer support costs by 35%
• Improved response relevance by 25% and enabled 120+ customer intent types
• Implemented explainability and compliance tooling with SHAP and LIME for regulated environments
• Designed scalable real-time ML infrastructure with Kafka, FastAPI, Docker, SageMaker, and Kubernetes
• Productionized adaptive fraud detection with drift monitoring and automated retraining
• Reduced false positive fraud alerts by 28% and cut fraud losses by 20-25%
• Standardized MLOps pipelines, reducing deployment errors by 40%
• Accelerated model refresh cadence from monthly to weekly

Experience

Education

Skills

Languages