Forward Deployed Engineer, Data De-Identification (Co-op)

Role overview

Deliver data de-identification projects from raw customer exports through quality-checked datasets ready for client acceptance. You will work hands-on with data pipelines, investigate processing issues, and improve delivery tooling so sensitive business records can be prepared for AI training while preserving their useful structure and context. This is a remote co-op role of approximately 20 hours per week, open to current students and experienced engineers seeking part-time work.

Responsibilities

- Run customer datasets through de-identification workflows and inspect results for missed entities, incorrect substitutions, inconsistent identities, and files that fail to rebuild. - Diagnose pipeline problems, fix them where possible, and clearly communicate issues that require help from other engineers. - Build or adapt ingestion and transformation steps for exports such as email, chat, documents, tickets, and databases. - Add code to the existing platform for reusable improvements as well as customer-specific data handling. - Reconcile file counts and content across ingestion and delivery, and coordinate quality checks with annotation colleagues. - Contribute technical details to client reports, including sensitive data handling, file processing, and known limitations; explain the work and results to clients. - Turn project-specific fixes into documentation, runbooks, and product improvements.

Requirements

- A technical bachelor’s degree or current enrollment in a relevant field such as computer science, engineering, mathematics, statistics, physics, or bioinformatics. - Practical experience in data engineering, backend engineering, or customer-facing solutions work that involved shipping systems; internships, co-ops, research, and substantial projects count. - Strong Python and SQL skills, familiarity with other backend languages, and confidence reading and debugging existing code. - Experience building repeatable pipelines for large structured, semi-structured, or unstructured datasets, with the ability to detect incorrect output. - A careful approach to AI-assisted work: know where tools and agents help, where oversight is needed, and verify their results. - Self-direction, clear communication with technical and nontechnical audiences, and a record of delivering complex work on short timelines. - Availability for about 20 hours each week, with flexibility to keep customer projects on schedule.

Nice to have

- Experience with model inference, machine learning pipelines, or PyTorch; data masking, de-identification, ETL, or synthetic data tools; or work at an early-stage company or new product line. - Interest in AI training data and working with advanced AI labs.

Benefits and work setup

- Competitive compensation. The source describes close collaboration with technical founders and product and data science teams.