Research Internship – ML Validation & Benchmarking ? Fully remote | ⏳ 6 months | ? Autumn 2026, flexible start About Kernel Science and the role We are a team of scientists who love building solutions. We take on R&D; projects for biotech and pharma companies, research foundations, deep-tech ventures, and we build our own products and companies across food science, multi-omics, scientific software, and AI/ML. Kernel Science builds Corollary Labs, a platform that brings scientific models, tools and datasets together behind one API and a supervised research agent. We are looking for a Master’s student to evaluate the platform’s performance, validate its outputs and benchmark it against alternative models. This research will help us understand where the platform performs well, where it fails, and how it compares with alternatives, turning results into evidence we can publish and act on. Responsibilities
• Design benchmark suites with ground truth, covering structure prediction, docking, sequence and molecular property tasks, as well as end-to-end research questions.
• Run Corollary Labs, competing models and general-purpose AI assistants on the same tasks, ensuring results are collected reproducibly.
• Analyse results statistically, using calibration, confidence intervals, error attribution and rigorous comparisons.
• Develop ML-based validation methods, including automated scoring of agent outputs and detection of systematic failure modes.




• Document findings in internal reports and, where results warrant it, contribute to a technical note or research paper with the team. Qualifications
• Currently enrolled in a Master’s programme in computer science, data science, statistics, physics, bioinformatics or a related field.
• Comfortable working with Python and Git-based codebases.
• A solid foundation in statistics and working knowledge of machine learning.
• Curiosity about how scientific models fail, combined with a rigorous and critical approach to evaluating results and claims.
• Working proficiency in English. Italian is welcome but not required. What you will gain
• The opportunity to investigate a real research question on a live product, with findings that can influence what we ship.
• Direct mentoring from the founding team and weekly reviews of your work.
• The opportunity to contribute to research outputs, including a technical note or paper where the results warrant it.
• The possibility of joining the team after the internship. Practical details The six-month internship is fully remote and can be curricular or extracurricular, subject to university agreement. The start date is flexible and can be agreed upon based on your university calendar.

How to apply
? Send your CV and a few lines about a model, dataset or analysis you have worked with, and what you learned from it. A link to code you have written is welcome.

📌 Stagista in ricerca e sviluppo (Italia)
🏢 Kernel Science
📍 Italia