Internship

Birlasoft Ltd

AI/ML Intern | June – July 2026

Worked with Birlasoft's CTO team to research and prototype next-generation AI systems for enterprise applications. My work centered on AG-UI, AI agent workflows, real-time streaming, human-in-the-loop interactions, and evaluating emerging AI protocols such as MCP, LangGraph, and A2A while building a Proposal Assistant prototype.

AG-UILLMsMCPLangGraphA2AGenerative AISSE

Context

The internship focused on understanding how enterprise AI applications move beyond simple chat interfaces. I researched modern AI communication protocols, studied agent orchestration concepts, and contributed to a Proposal Assistant prototype that streamed intermediate workflow updates while processing RFP documents. Alongside development, I evaluated how these technologies could be adopted within enterprise software.

Architecture

Researched the AG-UI protocol and its event-driven communication model.

Designed AI workflows around structured events, streaming updates, and human approval checkpoints.

Implemented real-time communication between AI agents and the frontend using Server-Sent Events.

Built the prototype using Next.js, Node.js and TypeScript while keeping the architecture modular for future AI workflows.

Birlasoft Logo

Birlasoft Ltd

Engineering Contributions

Core Systems

  • • AG-UI protocol research
  • • AG-UI event-driven backend
  • • MCP & LangGraph evaluation
  • • Reusable agent architecture
  • • Enterprise AI documentation
  • • A2A architecture exploration

Engineering Focus

  • • Enterprise Generative AI research
  • • Human-in-the-loop workflows
  • • Frontend-backend integration
  • • Technical documentation

Highlights

15+

AG-UI event types researched and evaluated

4

Enterprise AI technologies evaluated

3

AI courses completed

Reflection

This internship fundamentally changed how I think about AI applications. Rather than viewing LLMs as standalone models, I learned how enterprise AI systems rely on communication protocols, structured workflows, streaming, human oversight, and modular agent architectures. It gave me hands-on exposure to the emerging technologies shaping production AI systems.