Careers

Build the design layer for immunology.

Small team. Real problem. Open roles in San Francisco, hybrid.

  1. Member of Technical Staff

    Alignment & Reinforcement Learning

    San Francisco · Hybrid Full-time

    Up to $250K + meaningful equity

    Optimization under a hard constraint: reward is scarce, expensive, and arrives late. We are looking for someone who has done reinforcement learning in the regime where you cannot simply sample more.

    What you'll do

    • Own the RL formulation of therapeutic design — policy, reward model, and how a limited evaluation budget gets spent.
    • Build offline and off-policy methods that extract everything available from data already in hand.
    • Design the acquisition and uncertainty machinery that decides what is worth evaluating next.
    • Turn sparse, noisy, delayed signal into rewards that are worth optimizing against.
    • Keep the objective honest — retrospective benchmarks, ablations, and holdouts that predict what the next round will do.

    What we're looking for

    • A PhD in a related field, first-author publications at NeurIPS, ICML, or ICLR, or a body of work that speaks for itself. We care about the evidence, not the credential.
    • Depth in RL where samples are expensive: offline RL, off-policy correction, model-based RL, or Bayesian and active experimental design.
    • Hands-on with RL for small molecule design — policy-gradient generative models (REINVENT-style), GFlowNets, or comparable — and clear-eyed about where they break.
    • Treats reward hacking, distribution shift, and proxy-model collapse as daily engineering problems, not paper topics.
    • Strong PyTorch and the discipline to make a result reproducible.
  2. Member of Technical Staff

    Machine Learning

    San Francisco · Hybrid Full-time

    Up to $250K + meaningful equity

    Own how our models get trained. Pretraining and post-training at scale, multimodal architectures that hold heterogeneous biological data in one representation, and the efficiency work that sets how many ideas we can test in a month.

    What you'll do

    • Train and post-train foundation models over heterogeneous, multimodal data.
    • Own training efficiency end to end: parallelism, memory, throughput, and the cost of an experiment.
    • Adapt large pretrained models to domains where labeled data is scarce.
    • Build evaluation that tells us early whether a run is worth finishing.
    • Keep the training stack boring and reproducible, so results survive contact with the next model.

    What we're looking for

    • Demonstrated work training models at scale. A PhD is welcome but not required — a track record is.
    • Fluency in distributed training: FSDP or DeepSpeed, mixed precision, parameter-efficient adaptation.
    • Multimodal modeling experience, especially fusing modalities that share no natural tokenization.
    • Strong PyTorch, and the habit of profiling before optimizing.
    • Bonus: publications, notable open-source work, or a model other people actually use.
  3. Member of Technical Staff

    Computational Biology

    San Francisco · Hybrid Full-time

    Up to $250K + meaningful equity

    Drive the biology side of the foundation model. Architect the path from raw single-cell, TCR-seq, and multi-omics data to model-ready substrates, and design the experiments that tell us whether the model has it right.

    What you'll do

    • Design and analyze experiments that train and evaluate the platform.
    • Build computational pipelines for single-cell, TCR-seq, and multi-omics data.
    • Collaborate with the ML and experimental teams on data acquisition strategy and prioritization.
    • Publish and represent the science at major conferences.

    What we're looking for

    • PhD in computational biology, immunology, bioinformatics, or a related field.
    • Deep experience with single-cell analysis (scRNA-seq, scATAC-seq, scTCR-seq).
    • Strong Python and modern ML fundamentals.
    • Bonus: hands-on with CAR-T, TCR engineering, or tumor immunology.
  4. Clinical Partner

    San Francisco · Hybrid Full-time

    Bridge the platform with clinical reality. Translate model outputs into trial design, regulatory strategy, and the academic-medical-center relationships that make first-in-human work happen.

    What you'll do

    • Lead clinical strategy for Baseform's first CAR-T programs.
    • Build and steward relationships with academic medical centers, sponsors, and key opinion leaders in immuno-oncology.
    • Translate platform outputs into trial designs and IND-enabling strategies.
    • Partner with the science team on biomarker selection, patient stratification, and endpoint definition.

    What we're looking for

    • MD or MD/PhD with experience in cell therapy or immuno-oncology trials.
    • 5+ years of clinical development experience, ideally in oncology.
    • Direct experience with FDA interactions and IND filings.
    • Comfortable operating in early-stage ambiguity, with strong taste for what matters first.

Don't see a role that fits, but think you should be here?