Summer Intern

Job descriptionAt Pareto Labs, we’re building systems that make AI actually reliable in the real world.That starts with one thing: understanding where frontier models break — and fixing it.We’re opening a small Summer Research Program for people who want to go beyond using AI, and instead push it forward.What You’ll Work OnIdentify failure modes in frontier models (reasoning gaps, ambiguity, tradeoffs)Design hard, real-world scenarios where models struggleRun structured red teaming across LLMs and agent systemsTurn failures into:evaluation datasetstraining signals (RLHF / expert data)improvement loopsHelp build systems that make models more robust, aligned, and usefulThis is not just analysis.You will directly contribute to making models better.Who We’re Looking ForDeep curiosity about how and why AI systems failStrong problem-solving ability in messy, ambiguous environmentsBackground in:AI / ML / LLMsOR a domain where judgment matters (law, finance, ops, etc.)Ability to think beyond benchmarks into real-world complexityBonus:Experience with RLHF, evals, or adversarial testingExperience designing difficult prompts or simulationsWhy This MattersFrontier models are powerful — but still fragile.They fail when:There’s no clear “right” answerTradeoffs matterContext is incompleteFixing this layer is what will unlock real AI adoption.That’s what we’re building.Program DetailsType: Summer ResearchLocation: RemoteDuration: Flexible (8–12 weeks typical)Commitment: Flexible, outcome-drivenTop researchers may transition into core team roles.How to Applysend a short prompt where a frontier model fails + what the correct answer should be, then apply via LinkedIn.