SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Apheris is building AI infrastructure for pharmaceutical R&D, enabling leading pharma teams to discover and develop drugs faster. The company hosts federated data networks for drug discovery AI spanning co-folding, ADMET, and antibody developability. Models are trained on proprietary industry datasets to achieve superior performance while keeping data control and IP protected.
You will join the large molecule ML team as a hands-on ML engineer at the intersection of foundation models, structural biology, protein engineering, and federated learning. This is a substantial individual contributor role where you will own end-to-end model programs—taking research-led or open-source prototypes and turning them into production models that can be evaluated, released, and used in real drug discovery workflows.
Key responsibilities:
- Build, fine-tune, and extend large biomolecular models such as OpenFold, Boltz-2, and ESM for antibody modeling, co-folding, binder prediction, and developability
- Transform research code and prototypes into reliable components that run in federated training and evaluation pipelines
- Design evaluations and benchmarks, and deliver results packages for consortium partners
- Own workstreams through to release against agreed milestones, raising risks and trade-offs early
- Collaborate with product, engineering, research, and consortium members to ensure model work meets real application needs
You are an ML engineer who works close to the science. You have trained and evaluated models on biological data and can take a paper or open-source model and make it work on a new problem. You care about whether a model actually holds up in practice, not just whether the numbers look good, and you want your work to end up in real use by pharma R&D teams.
REQUIREMENTS:
Must-have:
- MSc, PhD, or equivalent experience in machine learning, computational biology, bioinformatics, physics, or a related field
- Strong Python and PyTorch; hands-on experience training or fine-tuning deep learning models on biomolecular data
- Hands-on experience with co-folding models or protein language models (OpenFold, AlphaFold, Boltz, ESM, or similar) beyond just running inference
- Good evaluation habits and solid engineering practice: fair benchmarks, reproducible experiments, and code others can build on
Nice-to-have:
- Experience with Kubernetes-based training, evaluation, or deployment, or other MLOps and ML infrastructure tooling
- Experience with federated learning, privacy-preserving ML, or distributed and multi-GPU training
- Experience in pharma, biotech, or other regulated or high-trust environments
- Publications in ML, computational biology, or structural biology venues (NeurIPS, ICML, ICLR, or similar)
