Sri Lekkha Sakhamuri

Ai Ml Engineer @Mizuho

US

Directed end-to-end development of a predictive credit-risk scoring platform, collaborating with product, data, and compliance teams. Integrated an explainable-AI module using LLM embeddings and prompt engineering to improve transparency and analyst productivity.• Engineered advanced feature sets from demographic, financial, and unstructured text data using Python, SQL, and transformer-based encoders, enhancing model accuracy by 12% and reducing bias through retrieval-augmented enrichment pipelines.• Designed and deployed ensemble models combining XGBoost, deep learning, and fine-tuned LLMs for tabular-text hybrid prediction, achieving a 15% AUC lift over baseline approaches.• Built RESTful scoring and explanation APIs (FastAPI) and implemented CI/CD pipelines with SageMaker, MLflow, and Docker to deliver low-latency, scalable inference services.• Optimized inference by applying model-compression techniques (quantization, ONNX) and distributed serving on AWS, reducing latency by 25% and supporting real-time credit decisions.• Established continuous-monitoring workflows (Prometheus, Grafana) to track drift, latency, and compliance metrics. Automated retraining pipelines to sustain robustness against changing financial and textual trends

Velagapudi Ramakrishna Siddhartha Engineering College

Bachelor of Technology - BTech, Computer Science

Narayana Junior College, Vijayawada

Intermediate, MPC

Ravindra Bharathi Public School, Vijayawada

SSC

Auburn University at Montgomery

Master's degree, Computer Science