Winter internship · Pre-final year · Applications open
Solumn AI Winter Internship
Evaluating the safety of frontier AI systems.
Six weeks on live AI-safety, cyber-security and model-alignment work, with a pre-placement offer on merit at the close.
The work
Six weeks on one live deliverable.
You own one live deliverable, start to finish. Nothing here is a teaching exercise: what you build is reviewed, graded and shipped — the evaluations a frontier lab clears before it ships.
AI safety evaluation
Tool-use and MCP safety evaluations that measure how an agent behaves under realistic attack conditions.
Cyber security
Vulnerability find-and-fix tasks built from real CVEs, backed by executable verification.
Model alignment
Agentic environments and deterministic graders for how a model holds the line under pressure.
What you walk away with
Real work, years early.
You build the evaluations a frontier lab clears before it ships — work most engineers encounter years into a career, if at all.
A pre-placement offer
Extended on merit at the close, at compensation benchmarked to frontier AI research roles.
The craft, early
Reviewed work, direct feedback and real depth in agentic safety, security and alignment.
Paid, and remote
₹1,00,000 for the six weeks, worked remotely, full-time over the winter.
Who can apply
Pre-final-year students at the IITs.
B.Tech · Dual Degree · M.Tech — CSE, AI/ML, Data Science, ECE, EE, Mathematics & Computing
Confident you can contribute in this space? Apply regardless of branch.
Selection process
Apply.
A few details and your CV. We review on a rolling basis and reach out to shortlisted candidates ahead of the start.