Work
I build AI pipelines, agent workflows, and retrieval systems that connect language models to enterprise data and tools.
Generative AI pipelines and agent harnesses
I work with clients to identify technical bottlenecks, design AI solutions, and build data pipelines and agent harnesses with observability to monitor model outputs, tool calls, and failures.
My work includes generative AI use cases and agent workflows using Snowflake Cortex for LLM calls. I also investigate output reliability issues, including unwanted process narration appearing instead of the requested result.
I also develop reusable generative AI packages that integrate into existing business workflows and enrich their outputs with AI capabilities.
Multi-agent orchestration and tool integration
Built configurable workflows that coordinate specialist agents across multiple steps. Integrated tools through MCP, HTTP, and native interfaces, and added execution traces to help debug model calls, tool calls, and changes in workflow state.
Retrieval and natural-language search
Built RAG pipelines with embeddings and vector search at Wipro. At a Hitachi Group company, built a backend that connected generative AI services to search and recommendation APIs.