- Built Domain Agents and RAG pipelines for multi-step reasoning workflows. Integrated LLMs with external tools and APIs for agentic task execution
- Implemented Setting as Code (EaC) practices to automate dev, staging, and prod environments for consistent ML deployments
- Developed CI/CD pipelines using Bitbucket, Jenkins, and Spinnaker for automated build, test, and deploy to AWS EKS
- Managed artifact lifecycle using Artifactory PyPI and Container Registry for versioned, secure ML model distribution
- Deployed scalable inference services using AWS AppConfig, Docker, and Kubernetes (EKS). Used VS Code and Devshell for cloud-native development
- Streamed real-time data using Kafka for low-latency inference and agent coordination
- Implemented Galileo Assurance for model evaluation, explainability (SHAP/LIME), and performance monitoring of LLM agents
- Integrated SmartSDK for agent development and ForgeRock for secure authentication and authorization in AI workflows
- Built observability and monitoring (AOM) pipelines to trace agent reasoning, token usage, and latency using MLflow and CloudWatch
- Automated deployments with Jules and managed infrastructure with Git-based workflows in Agile teams