Bioengineer, Lead Scientist
Listed on 2026-10-18
-
Business
AI Evaluation
Overview
Mission: Measure and improve AI model performance in biology by engineering benchmarks and leading a team to deliver for frontier model labs.
As a lead scientist, you will sit at the heart of Latch Bio's benchmarking team and report directly to our CTO. You will work closely with the therapeutics and drug development team, product engineering, and scientific leadership to build and scale benchmark and task production.
ResponsibilitiesBenchmark co-design: Map the "space" of your domain knowledge into a taxonomy of benchmarks, and co-author and release them for frontier labs to monitor and improve model performance. See (Use the "Apply for this Job" box below). for examples.
Hands-on task design: Design the tasks and graders that compose benchmarks yourself; set task standards and rubric quality for the team. Essentially, design the standards for what great is.
Production and quality: Once the team is in place, ensure the program meets volume and timelines agreed on with our partners. Volume is set per partnership and depends on its scope. The type of delivery may change by contract, eg. from eval creation to rubrics or preference labeling.
Quality Review
:
Review the team's work before delivery. Beyond automated platform checks, this means you need strong judgement of scientific value, clear and professional writing, tasks that fit the contract's scope, and decision‐making on whether a task teaches a real, transferable skill. You will also design and decide on quality and review standards for different types of work.Team mentorship & leadership: Lead a team of up to 7 scientists. Coordinate onboarding, 1:1s, feedback, and professional development. Be a reliable and communicative point of contact when they are blocked, technically or professionally. Advocate for team needs to leadership.
Operations support: Be a force multiplier:
Work closely with the engineering team to optimize compute and platform usage so individual scientists can focus on the science.Tools and product engineering: With the engineering team, build tools that automate tedious parts of task creation, and act as a core user who gives feedback on new platform tools.
Strategy
:
Lead domain initiatives such as data acquisition and roadmap definition, and help hire and close strong candidates. Communicate strategic initiatives directly with leadership to build consensus and execute across the organization.
Your rough milestones. Much of this work runs in parallel, so your timing will vary.
Month 1
Deliverable:
Meet the team and onboard to internal tooling. Design the structure of your first benchmark, in line with CTO and MTS-level thinking, and start drafting the first tasks. Support:
Scientific and strategy leads (MTS)
Month 2
Deliverable:
Design and launch a benchmark. Begin to manage and onboard first L1 hires on the team. Support: COO and recruiting leads
Month 3
Deliverable:
With the MTS in place, engage a frontier lab partner to scale up work and start hiring to build out the function. Support:
Head of Product and production leads
This is a normal full‑time role: about 40 hours per week. Nights and weekends are not expected. When crunches have happened (for example a first major contract), we adjusted quotas. The role is hybrid. Daily onsite presence is not required (though welcomed and encouraged), and some L2s are fully remote.
Your role vs. the MTS roleBoth roles work on and co‑author gene therapy benchmarks. The difference is ownership: the MTS owns the entire business vertical - hiring, production, deliverables, timelines, and customer relationships - while you own hands‑on management, scientific excellence and judgement, and tooling people rely on.
Benchmark…
(If this job is in fact in your jurisdiction, then you may be using a Proxy or VPN to access this site, and to progress further, you should change your connectivity to another mobile device or PC).