Ragesh Shanmughan
Ai Ml Solution Designer @Tata Consultancy Services
Signup · Get unlimited contacts
WORK HISTORY
Ai Ml Solution Designer @Tata Consultancy Services
Coimbatore, IN
Developed and deployed transformer-based models (PyTorch) for image classification, covering end-to-end architecture design, training, and production deployment.Implemented Human-in-the-Loop (HITL) workflows with configurable confidence thresholds and quality gates.Designed and implemented MCP-based integrations with Jira, ServiceNow, PagerDuty, and Slack with graceful fallback chains.Created multi-tenant architecture with tenant isolation, token metering, and cost management.Designed APIs and integration frameworks for seamless multi-agents AI deployment within client ecosystems.Implemented Agentic AI solutions using OpenAI, LangChain/LangGraph, and deployed via MLFlow & FastAPI for scalable AI workflows.
EDUCATION
Sahrdaya College of Engineering and Technology
Bachelor’s Degree, Computer Science
SKILLS
ABOUT RAGESH SHANMUGHAN
An IT professional with 17+ years of rich experience in designing and implementing AI/ML, agentic, cloud, IoT, infrastructure, compute, storage, networking, and developing multiple IT systems.• Highly motivated and results-oriented AI/ML Architect with a proven track record of success in building and deploying cutting-edge AI/ML solutions• Designed and implemented multi-agent workflow engine using LangGraph with parallel execution, achieving 40% faster ticket processing• Built RAG pipeline with GraphRAG enrichment, cross-encoder reranking, and reflexion loops for high-quality response generation• Integrated Agent Lightning RL for continuous learning from ticket outcomes, improving routing accuracy by 25%• Implemented Human-in-the-Loop (HITL) workflows with configurable confidence thresholds and quality gates• Developed MCP-based integrations with Jira, ServiceNow, PagerDuty, and Slack with graceful fallback chains• Cloud & MLOps: Extensive experience with AWS (EKS, SageMaker HyperPod, Bedrock), Kubernetes, and CI/CD automation (GitHub Actions, Terraform)• Full-Stack AI Implementation: From model development (PyTorch) to production deployment (MLFlow, FastAPI,MCP) including RAG systems and multi-agent architectures• Designed and implemented multi-agent workflow engine using LangGraph with parallel execution, achieving 40% faster ticket processing• Built RAG pipeline with GraphRAG enrichment, cross-encoder reranking, and reflection loops for high-quality response generation• Integrated Agent Lightning RL for continuous learning from ticket outcomes, improving routing accuracy by 25%• Established robust LLMOps/MLOps practices for reliable model lifecycle management• Proficiency in NVIDIA multi-GPU architecture, setup, training, and inferencing.• Hands-on experience with Nvidia Nim, Nemo, Megatron-LM, Tensor-RT, and Dynamo Services.• Configured and executed distributed training methods (Tensor, Pipeline Parallelism, Zero Redundancy Optimization) using a multiple-GPU approach.• Expertise and proficiency in fine-tuning techniques for LLM, such as SFT, LoRA, and DPO.• Created agentic AI technologies using OpenAI and Agent SDK to construct multi-agentic architecture.• Amazon Comprehend and Amazon Textract, to implement and optimize natural language processing and document extraction solutions.•Well-known in GenAI prompting/context engineering techniques and RAG technologies. • Highly skilled in the design and implementation of LLMOps, AgentOps, and MLOps pipelines.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.