Biotech Employer Shortlist soup-side

1. Where you actually fit

You get hired into software/infra seats, not science seats. Four archetypes recur across all five scouts:

• ML training-platform / MLOps — distributed training of protein/cell foundation models
• Inference & runtime perf — serving + optimizing those models (your low-latency lane)
• Bio-data platform / dev-infra — the data and compute layer under research
• Modeling & simulation engineer — stochastic biological models, self-driving-lab schedulers, drivers

Honest read: the sector runs Python/PyTorch/JAX/K8s. Nobody filters on Rust. Your Rust depth is a perf/systems differentiator you demonstrate, not a checkbox they match. What opens doors is distributed training/inference infra plus demonstrated systems work. A benchmarked standalone crate is evidence a hiring manager reads in five minutes.

2. Top shortlist (ranked by genuine fit)

1. Ginkgo Bioworks — Boston

• Does: Cell-programming foundry; the Ginkgo Datapoints / automation arm runs robotic labs at scale. Autonomous-lab scheduling and instrument drivers are core eng.
• Role: Platform/automation SWE — lab scheduling, device integration, control plane. Scheduling + drivers is a distributed/low-latency systems seat.
• Why fits: Autonomous-lab scheduling and drivers map onto your distributed-systems and low-latency background. Demonstrated systems work is the hiring signal.
• Loc/stage: Boston (Seaport), US. Public (DNA). Cut ~35% staff in 2025 — confirm the automation/platform org is hiring, not contracting.
• Apply: ginkgobioworks.com/careers

2. Asimov — Boston

• Does: Genetic-design foundry — builds models and tooling for designing biological systems (cell engineering, mammalian expression). Modeling and simulation is a named function.
• Role: Modeling & Simulation Engineer · Platform SWE. The M&S seat is the closest direct match to a stochastic gene-circuit / simulation wedge.
• Why fits: Your stochastic-modeling story (Gillespie/SSA, chemical master equation) lands on the modeling-and-simulation role, not a generic infra seat. The wedge matches the job description.
• Loc/stage: Boston, on-site. Backed by established synbio investors. Bends toward Python for biological-data modeling — bring the sim/perf angle.
• Apply: asimov.com/careers

3. Arc Institute — Palo Alto

• Does: Nonprofit AI×biology. Virtual Cell Initiative = full-stack learned model of the cell; scBaseCount curates all public single-cell data.
• Role: Infrastructure Engineer · Full Stack Engineer · ML Engineer (foundation model for perturbation response, $164–235k).
• Why fits: Virtual-cell infrastructure is a distributed-systems and simulation seat. Cell models are dynamical systems, so the low-latency simulation angle differentiates here.
• Loc/stage: On-site Palo Alto, US. $650M+ committed, Stanford/Berkeley/UCSF. Very stable.
• Apply: Greenhouse · Infra Eng req · Virtual Cell Initiative

4. Isomorphic Labs — London

• Does: Alphabet/DeepMind spinout, AlphaFold lineage, computational-first drug discovery. Engineering is the product.
• Role: SWE (Training Platform) / Lead SWE (Inference Platform) / SWE (Data Services) / Principal SWE (reliability+scale). Training/inference runtime, not glue.
• Why fits: Perf-critical platform roles = your Rust/low-latency depth translated to GPU/TPU infra. Soup-side mission (biology as learned distribution).
• Loc/visa/stage: London, hybrid 3d/wk. Sponsors UK visa + family relocation (verified). Alphabet subsidiary + $600M round — lowest funding risk on the list.
• Apply: Greenhouse · Job openings

5. Insitro — South San Francisco

• Does: ML + data-at-scale drug discovery (Daphne Koller). Petabyte cellular-data platform; non-hierarchical, eng-respected.
• Role: Senior/Staff SWE, Compute Core ($180–230k) — cloud infra, batch compute, observability. Senior SWE Data Platform. Senior/Staff Research Eng, HPC ($185–225k).
• Why fits: Compute Core + HPC are a systems-eng bullseye. Top comp (ML eng avg ~$337k per Indeed).
• Loc/stage: South SF, on-site. Active hiring.
• Apply: insitro.com/jobs · a16z board

6. Profluent — Emeryville, CA

• Does: AI-first protein design; ProGen3 frontier protein LM ("scaling laws for biology"), E1 retrieval-augmented PLMs.
• Role: ML Platform/MLOps · Senior SWE Data Platform · SWE, ML Infrastructure ($170–230k).
• Why fits: Owns full ML infra lifecycle (training, pipelines, inference). Systems seat where a benchmarked Rust artifact reads as proof.
• Loc/stage: Hybrid Emeryville 2–3d, no remote/EU. ~46 open roles. Backed by Spark/Insight/Air Street.
• Apply: Greenhouse · SWE, ML Infra

7. Recursion — Salt Lake City / London / Toronto

• Does: Public clinical-stage TechBio, "Recursion OS" drug platform. Large mature eng org.
• Role: SWE — Orchestration Platform (distributed compute, data pipelines) · Senior MLOps Engineer · EM ML.
• Why fits: Distributed-compute platform role; historically has had Rust/perf work. Big org = more layers but SWE valued.
• Loc/stage: SLC primary, London/Toronto options. Very active, multi-geo.
• Apply: Greenhouse

8. Chai Discovery — San Francisco

• Does: AI-for-biology lab (Chai-1/2 structure prediction).
• Role: SWE, Infrastructure — resilient production systems, 24×7 observability, 0→1 and 1→n scale-ups, perf.
• Why fits: Pure systems seat; low-latency background is a direct plus.
• Loc/stage: SF. Backed by Thrive, General Catalyst, OpenAI, Dimension; Eli Lilly partner. ~11 roles.
• Apply: jobs.ashbyhq.com/chaidiscovery

9. EvolutionaryScale (CZI/Biohub) — Redwood City / NY

• Does: ESM3 multimodal protein LM. Acquired by Chan Zuckerberg Initiative.
• Role: Senior SWE, Data Infrastructure (NY ~$130–180k) + a perf/infra role explicitly optimizing ML inference stacks.
• Why fits: Inference-optimization is your perf lane verbatim.
• Loc/stage: Redwood City or NY, hybrid. Ex-Meta-AI founders, $142M seed, now CZI-backed (watch re-org churn).
• Apply: Greenhouse

10. Basecamp Research — London

• Does: Metagenomics — samples extreme environments, builds the largest proprietary sequence DB feeding protein FMs.
• Role: Senior Platform Engineer — owns hybrid K8s fleet (on-prem + AWS), inference serving, hardware/systems optimization. Near-perfect systems fit.
• Why fits: Pure platform/systems role; "data + infra under the models."
• Loc/visa/stage: On-site central London. UK sponsorship plausible (confirm with recruiter). Well-funded, data moat.
• Apply: jobs.ashbyhq.com/basecamp-research

11. Genesis Therapeutics — San Mateo, CA

• Does: GEMS platform (generative + physics models), runs 1,000+ GPU / 10,000s CPU clusters.
• Role: SWE Core Infrastructure / SWE ML Infrastructure (multi-cloud, Ray/K8s/Terraform, 1000+ GPU training).
• Why fits: GPU-cluster scale matches; stack Python/Ray, so pitch the perf and simulation angle. Remote ML-infra variant exists.
• Loc/stage: San Mateo. $300M+ total (a16z, NVIDIA, T. Rowe, Fidelity).
• Apply: SWE Core Infra (Menlo board)

3. The wedge — standalone artifact plus a simulation story

Two pieces carry the application. First, a standalone systems artifact: a solo-buildable Rust crate where the SIMD/cache/determinism work is the whole product, no biology and no co-builder required. The what-to-build page ranks these; the front-runner is a Rust QUBO/Ising solver (simulated-bifurcation, SIMD sparse matrix-vector inner loop; users in quant/DeFi for portfolio-under-cardinality and DEX arb-cycle detection). That crate answers "show me systems work" for Ginkgo, Asimov, and Arc.

Second, the career wedge in stochastic gene-circuit modeling and low-latency simulation: Gillespie/SSA, chemical master equation, fast dynamical-system simulation. That story bridges from systems work to biology and lands on the seats shaped for it.

Also in frame for the systems/simulation angle: Isomorphic and Insitro (perf-critical platform and HPC seats), plus Genesis Therapeutics (physics-model simulation at GPU-cluster scale).

4. EU / remote

Relocation, not remote, is the honest path — and London is the only EU node with depth + sponsorship + roles shaped like yours.

If remote-EU is a hard requirement, the bio-specific options dry up — better remote odds come from Rust-heavy infra/dev-tools shops adjacent to bio, which aren't bio.

5. Honest gaps

• Rust is never the requirement. Zero companies across all five scouts confirmed Rust in production. Sell Rust as proof you can build the perf-critical layer, not as a stack match. The standalone artifact is that proof.
• The "biology" gap is real. Foundries (Asimov, Ginkgo) and several FM shops bend toward Python/GraphQL biological-data work. The stochastic gene-circuit modeling and simulation story bridges it, but it stays a bridge — strongest at Asimov's modeling-and-simulation seat and Arc's virtual-cell infra.
• Market is thin in two spots: pure low-level Rust perf seats in bio (don't exist as advertised), and EU-remote (collapses). Synbio funding fell ~70% YoY in 2025; confirm the automation/platform org is hiring before banking on legacy foundry balance sheets (Ginkgo cut ~35% staff; Amyris bankrupt).
• The opening move: ship one self-contained Rust artifact a stranger uses on day one (QUBO/Ising solver leads — see what to build), and contribute to noodles first (PR #296 async indexed reader, then #104/#139 parallel BGZF). A merged PR plus a benchmarked solo crate is systems-work evidence legible in five minutes — the credential the market won't hand you.