Sujith Cholleti

AI/ML Engineer with 5+ years designing production machine learning, generative AI, and agentic systems across regulated enterprise environments spanning financial services, semiconductor, and IT services. Experienced building LLM applications, RAG pipelines, multi-agent workflows (LangGraph, MCP), vector search, and prompt engineering to deliver secure, governed AI automation for enterprise teams. Skilled across Python, PyTorch, TensorFlow, LangChain/LangGraph, MLflow, Spark, Databricks, and AWS/Azure/GCP for scalable, reliable model-lifecycle delivery in production cloud environments.

Adept at partnering with product, engineering, data science, and security stakeholders to operationalize AI with responsible governance, traceability, and monitoring for dependable enterprise adoption.

AI Engineer at Capital One (2024-06 – Present)

• Revolutionized distributed training frameworks to enhance edge inference capabilities, achieving a 99.99% reliability rate across multiple deployments while minimizing latency in user interactions.
• Spearheaded distributed AI initiatives leveraging diffusion models, resulting in two filed patents that advanced the state of the art in collaborative learning and performance optimization.
• Consolidated model architectures and optimization techniques, enabling the processing of large datasets with improved performance metrics, directly impacting customer engagement and satisfaction.
• Streamlined training methods and inference strategies, achieving a 40% reduction in benchmarking time while enhancing model accuracy and operational reliability across distributed systems.
• Designed agentic AI workflows with LangGraph, MCP, and AWS Bedrock, automating financial operations while improving governance, traceability, and analyst productivity.
• Built RAG services with LangChain, foundation models, and vector search, enabling secure enterprise knowledge retrieval for financial operations and analytics teams.
• Engineered Python microservices and REST APIs for model inference, integrating Docker, Kubernetes, and CI/CD pipelines to strengthen platform deployment reliability.
• Standardized MLflow governance workflows, monitoring controls, and validation checkpoints, improving audit readiness for AI features across regulated enterprise environments.

Machine Learning Engineer at Intel (2023-05 – 2024-05)

• Developed deep learning training pipelines with PyTorch, and distributed processing, supporting computer vision workloads across enterprise experimentation environments.
• Optimized model inference with ONNX Runtime, TensorRT, and OpenVINO, improving deployment efficiency across edge, cloud, and production serving targets.
• Automated data preparation and feature engineering with Python and Spark, strengthening dataset reliability, reproducibility, and readiness for machine learning experimentation.
• Validated object detection and image processing models with YOLO and OpenCV, improving production readiness through statistical evaluation and error analysis.

Junior Machine Learning Engineer at IBM (2020-05 – 2022-07)

• Architected deep learning architectures for on-device deployment, leading to two academic publications that showcased innovations in model efficiency and scalability in real-time applications.
• Optimized training fine-tuning processes by implementing hardware-aware training strategies, resulting in a 30% increase in model performance and reducing computational costs in resource-intensive environments.
• Engineered on-device inference systems utilizing on-device ML techniques, significantly advancing scientific discovery through experimental frameworks that optimized resource allocation and reduced latency in processing.
• Pioneered AI model development for privacy-preserving ML, contributing to open-source ML projects that achieved real-world impact by enhancing data security and user trust across various applications.
• Fine-tuned transformer models with Hugging Face, RoBERTa, and DistilBERT, supporting clause extraction and classification for legal and financial documents.
• Created semantic search capabilities with Sentence-BERT embeddings and FAISS indexing, improving document retrieval for compliance, policy, and risk management teams.
• Containerized machine learning services with TorchServe, ONNX Runtime, Docker, Kubernetes, and FastAPI, standardizing model serving across production workflows.
• Established validation and drift monitoring frameworks with Great Expectations, PSI checks, and MLflow controls, strengthening governance for enterprise machine learning releases.

Master's in Data Science – University of Maryland Baltimore County