About SnowRock
SnowRock is a data brokerage connecting U.S. business owners with buyers seeking data for AI training and evaluation. We research potential matches and coordinate introductions. Owners agree licenses and deliver data directly to buyers.
About this role
A useful data sample needs more than a readable file. A buyer may need to check whether records are complete, whether related documents can be joined and whether later information would accidentally reveal an answer the model should infer. You will turn these questions into small, repeatable technical checks.
You will work with research, sourcing and commercial colleagues to define evaluations and explain the findings. Owners license and deliver their data directly to buyers. Where a technical review is agreed, you will design tools that can run in an approved owner or buyer environment and clarify the access required. Your findings should help the buyer decide whether a larger evaluation is justified and explain what additional work would be needed.
This position is based in Michigan, with SnowRock locations in Ludington and Traverse City. We will discuss work arrangements during hiring.
What you will do
- Translate a buyer’s task into a test plan with explicit inputs, expected outputs, sample selection and a clear acceptance question.
- Build checks for file readability, field completeness, duplicate records, inconsistent units and links between records that should describe the same event.
- Design time-based splits and other controls that prevent future outcomes or repeated records from making an evaluation misleading.
- Create simple comparison methods so buyers can see whether added data changes performance relative to a relevant starting point.
- Record test configuration, model version, sample boundaries and failures so another reviewer can reproduce and interpret the result.
- Explain errors with concrete examples, distinguishing source-data limitations from parsing problems, task design and model behavior.
- Keep technical reviews within the agreed access and scope, and help both sides understand preparation work before committing to a larger evaluation.
What you bring
- Experience writing reliable Python and database queries, with examples of tests or analyses another person can reproduce.
- Ability to work with structured records and documents, investigate data quality problems and trace an unexpected result to its cause.
- Understanding of sampling, comparison methods and common ways evaluation results become misleading.
- Clear technical writing that describes what a test establishes and what would change the interpretation.
- Relevant engineering or research experience, independent projects or equivalent practical evidence; a particular degree is not required.
Additional experience we value
- Experience evaluating document understanding, retrieval, business reasoning or software-related AI tasks.
- Experience building checks that run within restricted customer environments.
- Experience estimating evaluation costs and reducing unnecessary model calls without weakening the comparison.
Compensation
The annual base salary range for this position is $270,000–$365,000 USD. Equity and benefits are additional. Your written offer will specify the package for your role.
Apply to SnowRock
Research Engineer, Data Evaluation
Email your resume or professional profile and a short note about relevant work to [email protected]. Work samples can be useful; share only material you have permission to disclose.
We welcome relevant experience gained through employment, independent projects, education or other routes. If you need an adjustment to take part in the hiring process, tell us when you apply.
Do not include identity documents, bank details or confidential information from a current or former employer. Read how we handle job application information.
Apply by email Opens your email app. Add your resume or work links before sending.