Internship: Evaluating Retrieval-Augmented Generation Systems

About the role

D ONE helps companies make sense of their data and generate value. This fixed-term internship (typically 3 months, 80-100%) in Zurich is open to students at Bachelor, Master or PhD level. You will work independently on a topic in evaluating generative AI systems, turning vague questions about quality into repeatable, trusted measurements.

Key responsibilities

• Review evaluation methodology for retrieval-augmented and generative systems.
• Design and implement automated evaluation logic in Python.
• Build and curate test data sets, including gold standards and synthetic cases.
• Analyse results statistically and turn them into clear, defensible conclusions.
• Document your work and present it to both technical and non-technical audiences.

Required profile

• Current enrolment in a Bachelor's, Master's or PhD programme in Computer Science, Data Science, Computational Linguistics, Engineering, Mathematics or a comparable field.
• Interest in at least one industry domain and curiosity about how data creates value in it.
• Pragmatic, results-oriented attitude and excellent English communication; German is a strong plus.
• Preferred: familiarity with LLMs and RAG, evaluation concepts (faithfulness, groundedness, hallucination detection), tools such as RAGAS, DeepEval, promptfoo, TruLens or LangSmith, LLM-as-a-judge techniques, and statistical foundations such as sampling and confidence intervals.

Required skills

• Experience with large data sets and databases.
• Knowledge of relevant technologies and programming languages (e.g. Python, SQL, Git, Spark, cloud platforms).
• Preferred: software engineering discipline (reusable Python packages, automated tests, CI, reproducible pipelines) and visualisation skills (pandas, plotly, matplotlib, dashboards).

What we offer

• A focused, low-noise environment and an international team.
• Mentoring by senior consultants and access to the D ONE Academy.
• A clearly scoped topic you can shape, frequent offsite events and a realistic path to becoming a consultant.