Data Science & GenAI Research Intern - IGEN

U.S. Venture, Inc.•Appleton, WI

2d•Onsite

About The Position

Join the Insight Engineering team for Summer 2026, starting May 2026. This internship is dedicated to applied research and development in data science and Generative AI — with a particular focus on agentic systems that reason over enterprise data. You will work alongside the foundation team that is standing up our Snowflake-anchored data platform, semantic layer, and KPI certification discipline. Your role is the R&D wedge: prototyping the AI capabilities that will eventually be embedded into IGEN products and shared as certified, governed insight with clients and partners. This is not a ticket-queue internship. You will own experiments end to end — framing the question, building the prototype, evaluating it honestly, and writing up what you learned so the team can decide what to productize. This internship will be located onsite at our Corporate Headquarters [222 W College Avenue, Appleton, WI 54911]. This intern will sit inside the Insight Engineering squad alongside the Sr. Data Engineer and BI Engineer, and will partner with product squads where R&D output intersects their roadmap.

Requirements

Pursuing a degree in Computer Science, Data Science, Statistics, Applied Math, Machine Learning, or a related quantitative field.
Working proficiency in Python, including common data and ML libraries (pandas, scikit-learn, NumPy).
Basic SQL fluency — you can write queries, read execution plans casually, and understand joins, aggregations, and CTEs.
Exposure to LLMs or GenAI tooling — coursework, side projects, or hands-on use of APIs (OpenAI, Anthropic, open-weight models, LangChain/LlamaIndex, or equivalents). You do not need production experience; you need genuine curiosity and a portfolio of attempts.
Comfort with ambiguity. R&D problems are not specced like feature tickets. You will need to scope your own work, decide when an experiment is done, and communicate trade-offs clearly.
Strong written communication. Findings the team can’t read are findings the team can’t use.

Nice To Haves

Hands-on experience with agentic frameworks (tool use, function calling, multi-step reasoning, ReAct-style loops).
Familiarity with retrieval-augmented generation (RAG), vector databases, embeddings, and evaluation methods for LLM outputs.
Exposure to Snowflake, dbt, or modern data warehouse tooling — Cortex AI experience is a clear plus.
Experience with Power BI or other BI/visualization tools.
MLOps awareness — experiment tracking (MLflow, Weights & Biases), model evaluation harnesses, reproducibility practices.
A public artifact — GitHub repo, blog post, paper, hackathon submission — that shows how you think when you build something.

Responsibilities

Prototype agents that answer business questions by reasoning over certified KPIs, semantic models, and operational data — not by guessing.
Build retrieval-augmented pipelines over structured warehouse data and unstructured documents (regulatory content, internal documentation, ticket history).
Explore tool-use patterns where an LLM orchestrates SQL generation, validation against the semantic layer, and result interpretation — with governance and lineage preserved.
Evaluate model behavior rigorously — accuracy, hallucination rates, latency, cost — and document trade-offs.
Develop and test models against governed datasets in Snowflake — forecasting, classification, anomaly detection, or entity resolution depending on the active research question.
Engineer features that could land in a future feature store, with attention to lineage, reproducibility, and certified inputs.
Run experiments using Snowflake Cortex AI (native LLM and ML functions in SQL), Python notebooks, and other appropriate tooling.
Write up findings in a way the engineering team, product partners, and SLT can consume — what was tried, what worked, what didn’t, and what to do next.
Demo prototypes to the Insight Engineering team and contribute to the decision of whether a capability is ready to move from R&D into the platform roadmap.