← Biotech Career Guide

Generative Biology

ML that designs biology — the field, the employers, and where a systems engineer fits · verified June 2026

One line
Generative biology designs biological matter — proteins, genomes, cells, molecules — instead of merely predicting it. The labs doing it are bottlenecked by engineering and compute, not ideas. That is the opening for a Rust/distributed-systems engineer.

The field, in five modalities

ModalityWhat it designsFlagship models (2025–26)Real now vs hype
Protein design Novel backbones + sequences, binders, mini-proteins RFdiffusion2/3, AlphaProteo, ESM-3, Chai-2, BindCraft Binders & mini-proteins real; general function-conditioned design still weak
Genome / sequence LMs DNA, variant effects, regulatory elements Evo 2 (Arc, 40B params, 1M-bp context, Nature 2025) Variant-effect prediction real; viable designed genomes still aspirational
Virtual cell Cell-state & perturbation/drug response Arc State, scGPT, CZI Virtual Cells Platform Works on seen cell types; causal generalization unsolved
Generative chemistry 3D small molecules fit to a binding pocket Pocket-conditioned diffusion (DrugDiff, fragment-based) Diverse candidates; synthesizability & scoring are the bottleneck
Antibody / enzyme CDRs, nanobodies, catalytic active sites RFdiffusion antibodies, Chai-2, RFdiffusion2 enzymes Zero-shot binders hit in a 24-well plate; de novo antibodies still hard

The common shape: a generative model proposes a sequence or structure, a predictor scores it, and a wet-lab round returns the truth. The model is the smallest part.

State of the art, 2026

What actually works at scale

De novo protein binders. The wet-lab data loop — not bigger models — lifted hit rates from under 1% to tens of percent. BindCraft (Nature 2025) averages ~46% (range 10–100%) at nanomolar affinity; AlphaProteo reports 9–88% across targets.
Antibodies / nanobodies. Chai-2 does zero-shot de novo antibody design at a 16–20% hit rate across 52 antigens with ~20 candidates per target — over 100× prior methods (which were under 0.1%). This is the year's headline jump.
Genome generation. Arc's Evo 2 generates genome-scale DNA and does zero-shot variant-effect prediction across all domains of life.

Still research: reliable novel enzyme catalysis; virtual-cell perturbation prediction on unseen cells/genes (the Virtual Cell Challenge drew 1,200+ teams — the signal that it is not solved).

Failure modes to know before the interview: data scarcity (perturbation and catalysis data are thin), the wet-lab validation gap (in-silico filters overstate success), and weak generalization to novel chemistry or cell types. Models shine on trained targets and falter off-distribution.

Where the money goes: binders and antibodies (clear clinical payoff, fast loops) attract capital today; the virtual cell is the next frontier bet (Arc, NVIDIA, 10x).

The companies (as employers)

⚙ = most engineering/compute-bound — these hire software/infra engineers, not only ML scientists. EU/UK flagged for the non-US path.

CompanyLocationGeneratesStage / fundingEng-bound
EvolutionaryScale (ESM)NYCProtein LLMs (ESM3)$142M seed (Amazon/Nvidia)⚙ frontier-model infra
Chai DiscoverySFAntibody/structure (Chai-2)$130M Series B, $1.3B val (OpenAI-backed)
Cradle BioAmsterdam + Zürich — EUProtein-design SaaS~$102M⚙ software-first
Latent LabsLondon + SF — EUDe novo proteins$50M (DeepMind/AlphaFold alumni)
Isomorphic LabsLondon — EUGenerative drug design$600M (Alphabet)⚙ ML research eng
Basecamp ResearchLondon — EUGenomic data + gene-insertion models~$85M (Nvidia)data/infra
Arc Institute (Evo)Palo AltoGenome/DNA-RNA-protein (Evo 2)Nonprofit, Nvidia compute⚙ compute-bound
Recursion (RXRX)Salt Lake City (public)Phenomics / cell modelsBioHive supercomputer⚙ large SWE org
Generate BiomedicinesSomerville MAProteins / antibody therapeutics$1B+ raisedSWE/MLOps
ProfluentBerkeley CAProteins, gene editors (recombinases)$150M + $2.2B Lilly pactbioinfra
Dyno TherapeuticsWatertown MAAAV gene-therapy capsidsSeries Amostly ML-science
Markov BioSFVirtual cell (aging/longevity)Early-stage, thin public data⚙ "limited by engineering & compute"

EU/UK without relocating to the US: Cradle (NL/CH), Latent Labs (London), Basecamp (London), Isomorphic (London). See also the Opportunity Catalogue (European hubs folded in) and the Markov Bio profile.

Where a systems engineer fits

The bottleneck is rarely the model — it is the loop around it. GPU memory and IO bandwidth, not algorithms, cap training and serving; only a handful of orgs control the data and GPU scale to build a foundation model. That is plumbing, scheduling, and throughput — your work, not a biologist's.

  1. Data pipelines at TB–PB scale. The canonical flow is FASTQ → align → BAM → count matrix → h5ad/AnnData. Training on the 100M-cell Tahoe-100M dataset already forces custom streaming loaders (e.g. annbatch) — IO, sharding, zero-copy. Reproducible Nextflow-style DAGs are the substrate.
  2. Lab-in-the-loop orchestration. Genentech/Prescient's LitL chains a generative model, a property predictor, active-learning selection, and in-vitro assays into a semi-autonomous loop — a distributed scheduling/queueing problem, not an ML one.
  3. DBTL automation. The Design-Build-Test-Learn cycle needs the glue that turns designs into lab runs and results back into training data ("Round Trip" systems).
  4. Serving + eval harnesses. Latency-bound inference servers for structure/sequence models, plus benchmark standardization — a named open gap.

What to build

The build path is its own page now — ranked, standalone artifacts a single engineer can ship (no agent-orchestration glue). See What To Build To Break In: a Rust QUBO/Ising solver leads, with a float compressor, a sparse direct solver, and bgzf-turbo behind it, plus the noodles contribution as the on-ramp.

Sources

  1. Evo 2 — Arc Institute (Nature 2025): arcinstitute.org/news/evo2
  2. BindCraft (Nature 2025): nature.com/articles/s41586-025-09429-6
  3. Chai-2 antibody design (bioRxiv 2025): biorxiv.org/.../2025.07.05.663018v1
  4. Arc State virtual-cell model: arcinstitute.org/news/virtual-cell-model-state
  5. Virtual Cell Challenge (Cell 2025): cell.com/.../S0092-8674(25)00675-0
  6. Generative protein-design review (arXiv): arxiv.org/pdf/2502.19173
  7. RFdiffusion functional-binder critique (bioRxiv 2025): biorxiv.org/.../2025.02.07.636769
  8. Genentech lab-in-the-loop (bioRxiv): biorxiv.org/.../2025.02.19.639050