Naga Venkata Padala

About

GenAI/ML engineer who has shipped production agentic systems in highly regulated and high-throughput environments, including an AWS Bedrock-based fraud/compliance workflow at U.S. Bank with PII redaction and hallucination detection that cut investigation time by 50%+. Also built and evaluated RAG and recommendation systems at Target, using RAGAS-driven testing, hybrid retrieval with re-ranking, and SHAP explainability dashboards to align model behavior with merchandising business KPIs.

Experience

Education

Key Strengths

• Deployed AWS Bedrock agentic system for fraud/compliance, cutting manual investigation time by 50%+
• Built compliance-grade GenAI safeguards (hallucination detection, PII redaction) for regulated banking use cases
• Designed low-latency real-time ML pipelines (Spark Structured Streaming sub-second scoring) at high volume
• Optimized model inference performance using ONNX and quantization
• Production MLOps/infra delivery with Terraform + FastAPI on Kubernetes with autoscaling
• End-to-end ML lifecycle automation with Airflow + MLflow (retraining, dependencies, versioning)
• Metric-driven agent/RAG evaluation using golden datasets and RAGAS to prevent regressions
• Operational monitoring for latency and concept drift using Prometheus/Grafana
• Strong retrieval strategy design (hybrid keyword+semantic search with re-ranking) to improve precision and control cost
• Effective cross-functional collaboration: aligned recommender optimization to business KPIs and delivered SHAP explainability dashboards