AI Research Intern

This internship points at the open problems rather than at a backlog: agentic systems, reinforcement learning for constrained design, ML-based physics models and the reliable generation of engineering artifacts. They are real research problems and they sit directly underneath a shipping open-source product, so a result that holds up ends up in something hardware engineers use rather than in a slide. You would work with the founder on what gets tried, own a direction rather than a set of isolated intern tasks and publish or open-source what comes out of it. Full-time and on-site in Bengaluru.

What you would own

• Research new approaches to schematic generation, component selection, placement and routing.
• Build and evaluate AI agents that plan, modify and verify PCB designs.
• Explore reinforcement learning for constrained design and optimisation problems.
• Develop ML-based physics and surrogate models for electrical and physical validation.
• Implement ideas from recent AI, ML, EDA and computational engineering research.
• Design experiments, evaluation datasets, benchmarks and reliability metrics.
• Integrate research prototypes with KiCad and deterministic engineering tools.
• Analyse failures, and turn the experiments that look promising into production systems.
• Document findings and contribute to technical reports or research publications.
• Contribute directly to copperhead’s open-source codebase.

What we look for

• Strong foundations in machine learning and deep learning.
• Proficiency in Python, and experience with PyTorch, JAX or a similar framework.
• Experience with LLMs, transformers, tool use or agentic systems.
• You can read a research paper, understand it and implement the idea.
• Familiarity with experimental design and quantitative evaluation.
• Strong software engineering and problem-solving skills.
• Curiosity about electronics, PCB design, EDA or computational engineering.
• You can work independently and take a research project from hypothesis to prototype.

Previous PCB design experience is helpful and not required. Evidence of curiosity, strong experiments and things you have built matters more than credentials.

Nice to have

None of this is required. It is what would make the work easier from the first day, not what decides the first email.

• Reinforcement learning or combinatorial optimisation
• Physics-informed machine learning or graph neural networks
• Simulation, surrogate modelling or scientific computing
• KiCad, circuit design or another EDA tool
• Publications, research projects or technically ambitious open-source work
• AI systems you have built that real users used