●​ Led the design and development of AI agents (Data Analysis, EDA, Financial Data Analysis, Custom GPTs) using AutoGen and LangGraph, which automated data workflows and reduced manual processing time by 60%, deepening my expertise in multi-agent orchestration and framework extensibility.

●​ Experimented, Built and deployed a Retrieval-Augmented Generation (RAG) techniques—both with and without agents—to optimize response accuracy; deployed a performant LLM-based chat engine and gained hands-on experience in tuning prompt pipelines and integrating semantic retrieval strategies.

●​ Created a lightweight, file-watcher-based centralized logging framework for Kubernetes-hosted apps, enabling real-time log aggregation without external tools; this improved debugging efficiency and taught me how to design minimal-observability systems tailored for constrained environments.

●​ Explored and integrated Model Context Protocol (MCP) to enhance coordination among AI agents, resulting in improved workflow cohesion and reduced integration overhead, while expanding my understanding of protocol-based interoperability in LLM systems.

●​ Coordinated with infrastructure teams to deploy applications in Kubernetes, achieving reliable and scalable hosting for AI services, and strengthening my ability to bridge development and DevOps practices for ML system deployment.

●​ Set the strategic direction for AI/ML initiatives within the team by introducing and driving adoption of emerging tools and architectures such as Model Context Protocol (MCP), multi-agent systems (AutoGen, LangGraph), and observability frameworks like Langfuse; enabled faster experimentation, improved system transparency, and aligned the team with state-of-the-art LLM development practices.

●​ Worked closely with cross-functional teams to deliver machine learning-based software solutions.