Infrastructure Engineer - LatchBio

We build the benchmarks that test whether frontier AI can actually do computational biology. Pharma, biotech, and solution providers rely on benchmarks.bio to evaluate and improve their AI systems - and we're just getting started.

We released the first rigorous benchmark suites for biology agents. They contain hundreds of real-world problems drawn directly from industry workflows. Our research shows that harness design - the tools, prompts, and control flow around a model - moves outcomes as much as swapping the base model itself. This insight is reshaping how frontier labs think about building with AI.

We've grown revenue 6x in the past 12 months. We have a repeatable sales process and deep partnerships with 10X Genomics, Vizgen, Takara Bio, and AtlasXOmics. More importantly: we have more demand than we can meet. That's why we're scaling rapidly across engineering. We're building benchmarks across omics, therapeutics, and surveillance. We need engineers to build the infrastructure that powers our benchmarking pipeline - data ingestion, eval design, grading systems, and infrastructure. Work with scientists and frontier models. Ship fast.

Here is a sample of engineering products we work on

• Agent harness integrated with bio first tooling such as an in-house h5ad and genomics viewer
• Distributed file-system with metadata on Postgres and blobs on S3
• Cluster orchestrator designed to accept compute nodes from anywhere on the internet
• Reactive Python-based notebooks for data analysis and visualization
• Benchmarking harness with automated QC processes to validate and probe model failure

cases

About this role

You will own the infrastructure layer of latch including container orchestration, distributed filesystems, cloud architecture, and database systems. While our team is operationally proficient, we do not have the bandwidth to tackle exciting areas of improvement such as speeding up container starts, improving filesystem throughput, scaling up RL operations, and reducing costs.

Signs of strong fit

• Worked both on and with container orchestration systems such as K8s
• Deeply proficient at postgres
• Mastery of cloud primitives, especially AWS ones
• Exceptional modern linux fundamentals like eBPF, cgroups2, etc

How we work

We care about output and let people optimize their work in a way that works for them. Our engineering team is in-person in San Francisco. We run a daily standup where engineers talk through technical challenges and stay aligned on what matters most. You will own a product area and execute, working closely with seasoned Latch engineers, product managers, and designers. Mentorship happens through code review and working through hard problems together.