Member of Technical Staff, AI Engineering

ABOUT HANDSHAKE Handshake's mission is to organize expert human knowledge to advance the AI economy. Handshake AI works directly with frontier labs on their most consequential data, evaluation, and post-training challenges, building the systems that turn expert human knowledge into the data and evaluations that make frontier models better. You will work alongside engineers, researchers, operators, and builders from organizations including Scale AI, Meta, Google, Amazon, xAI, Notion, and Palantir—and help build the systems that make expert human knowledge useful for advancing AI. ABOUT HANDSHAKE LABS Handshake Labs is building external AI products, research platforms, and customer-facing AI systems. We are evolving work that is often custom-built for an individual partner into reusable products and platforms that improve with every deployment. Our work spans the full post-training loop: designing evaluations and training environments, building high-quality data and feedback systems, running experiments, and turning what works into durable infrastructure. For example, we are developing agents that can analyze long, complex coding-agent sessions in days rather than weeks—with expert review and calibration built into the system. THE ROLE We are hiring a Member of Technical Staff to help build the data systems that make frontier model training possible. The data Handshake builds for and acquires on behalf of labs is getting more complex and more sensitive, and this role is responsible for improving how we generate, process, and prepare that data—whether that means building higher-quality synthetic and LLM-generated training data, or making acquired third-party data safe to use by removing personal information while preserving the structure that makes it valuable. You will partner with researchers, domain experts, legal/compliance stakeholders, and customers to turn ambiguous data questions—about generation, quality, evaluation, or privacy—into experiments, pipelines, and durable products. Early members of the team will have unusual influence over our technical direction, standards, and culture. Location: San Francisco & Mountain View preferred; open to exceptional candidates in other locations (London, Canada, Bangalore, etc.) WHAT YOU’LL DO - Design and build systems that improve the quality, scale, and safety of the data Handshake generates and acquires for frontier model training—spanning synthetic data generation and data anonymization/PII removal. - Translate ambiguous research, partner, or compliance needs into clear hypotheses, experiments, evaluation plans, and production-quality implementations. - Build and improve data-processing pipelines, evaluation frameworks, benchmarks, and quality-control systems, whether the goal is generating higher-signal synthetic data or verifying that sensitive data has been properly de-identified. - Run fast, rigorous iteration loops: prototype, evaluate, interpret results, and turn learnings into the next system or product. - Partner directly with researchers, domain experts, and—where relevant—legal and compliance teams to ensure data is both high-utility and responsibly handled. - Identify repeatable patterns across engagements and productize them into reusable software and platforms. - Raise the technical bar through strong design judgment, clear communication, code quality, and mentorship. WHAT WE’RE LOOKING FOR - 2–10 years of recent, demonstrated experience in one or more of: synthetic/LLM-generated data, post-training and model-evaluation work, privacy engineering, or data anonymization/de-identification at scale. - A hands-on individual contributor track record—this is not a team-lead or engineering-management role. - Strong Python skills and the ability to write clean, efficient, scalable software for large, messy, real-world datasets. - Sound judgment for reasoning about data quality, risk, and utility—forming hypotheses, choosing meaningful metrics, diagnosing failures, and distinguishing signal from noise. - Experience designing systems—not only implementing specifications—including tradeoffs around quality, scale, reliability, and reuse. - Comfort operating in an ambiguous, fast-moving environment with substantial ownership. - Collaborative, low-ego communication and the ability to work effectively with researchers, engineers, domain experts, and customers. ESPECIALLY COMPELLING EXPERIENCE - Building or operating large-scale synthetic or LLM-generated data pipelines for model training. - Building or operating large-scale data de-identification or anonymization systems, ideally involving relational or graph-structured data, with experience preserving referential/relationship integrity after anonymization. - Developing LLM/agent benchmarks, evaluation methodologies, annotation systems, or data-quality frameworks. - Research or applied work on reinforcement learning, align