Machine Learning Research Scientist (Co-Folding and Affinity)

Role Overview

Machine Learning Research Scientist specializing in Co-Folding and Affinity at SandboxAQ (full remote). You will help develop and iterate deep learning models for structure prediction and binding affinity, driving benchmarking, cross-team collaboration, and clear communication of results.

Key Missions

• Develop and iterate co-folding/core-replyment models by implementing, experimenting with, and refining deep learning approaches.
• Design and run systematic evaluation pipelines to measure model performance against state-of-the-art methods.
• Collaborate with multidisciplinary teams to integrate validated models into drug discovery workflows ready for production.

Responsibilities

• Execute rigorous controlled experiments and critically interpret results.
• Benchmark model performance vs. leading baselines and communicate findings.
• Support deployment of ML workflows into production-oriented discovery pipelines.

Requirements

• Ph.D. in Computational Biology, Biophysics, Computer Science, Computational Chemistry, or related field, with research focus on protein structure prediction and/or co-folding.
• Demonstrated ability to design controlled experiments and iterate effectively.
• Hands-on experience with protein structure prediction and/or protein-ligand co-folding, such as AlphaFold2/3, RoseTTAFold, Chai-1, OmegaFold, or comparable systems (developed via graduate/postdoctoral research).
• Experience developing, training, and validating deep learning models for structural biology, including familiarity with transformers, equivariant neural networks, and/or diffusion models.
• Strong Python skills and modern ML frameworks such as PyTorch and/or JAX.
• Strong written and verbal communication and ability to work in a fast-paced multidisciplinary research environment.
• Active or recently completed postdoctoral experience in co-folding, structure-based drug design, or closely related domains.
• Familiarity with binding affinity prediction, including structure-based or physics-informed approaches.
• Research output via publications/preprints in venues such as NeurIPS, ICML, Nature Methods, PLOS Computational Biology, or bioRxiv.
• Cloud computing experience deploying ML workflows on public cloud (e.g., GCP, AWS, or Azure).

Nice to Have

• Exposure to ML techniques for structural biology such as:
  • Generative models for protein/ligand design
  • Active learning for data generation
  • Foundation models for biomolecules
  • QSAR/property prediction
• Familiarity with drug discovery workflows (hit identification, lead optimization, SBDD).
• Experience with agentic coding tools (e.g., Claude Code, Codex) to accelerate prototyping.