AI & Agent Architect Intern

About this internship

1 About Power Flux AI

Power Flux AI is an AI-native power electronics engineering company headquartered in Pune, India. We design, simulate, prototype and productionize high-power motor and inverter systems, from 0.5 kW to 500 kW, on a proprietary AI-driven platform. It compresses traditional 1836 month development cycles into 90 days from concept to prototype.

We work with partners and institutions including IIT Bombay, IIT Madras and ARAI Pune, and our engineering is led by a founder with 35+ years of hands-on power electronics experience.

2 Your opportunity at IIT Madras Incubation, Chennai

Work with us at the IIT Madras Incubation ecosystem in Chennai, close to the Power Electronics Lab and IIT Madras researchers. You will build AI agents for a domain where mistakes are costly, so verification matters as much as generation.

3 Role overview

You will help build our agent orchestration layer: it takes a fuzzy circuit specification, splits it into engineering steps, runs them across the best available LLMs, checks results against datasheets and simulation, and produces verified designs.

4 What you will do

• Build parallel, DAG-based task execution for multi-step design workflows.
• Route tasks between reasoning-heavy and fast extraction models based on complexity.
• Build grounding pipelines using component datasheets, design constraints, past designs and simulation feedback.
• Write parsers that turn structured LLM outputs into typed, validated netlists.
• Design human approval checkpoints before irreversible actions.

5 What you bring

Must have

• Python with asyncio, type hints and pytest
• Hands-on use of LLM APIs, structured outputs and tool calling
• A project using RAG, embeddings or a vector store
• Understanding of agent loops and prompt design

Good to have

FastAPI, Pydantic, LangGraph or workflow engines, PDF parsing (PyMuPDF), compiler or AST basics, electrical engineering fundamentals.