5 months ago
San Francisco, CA, USA or New York, NY, USAMid Level
Responsibilities
- Build and maintain robust multi-stage asynchronous workflows for machine learning data generation, training, and evaluations.
- Design and rationalize machine learning systems and software architecture.
- Identify blockers and develop solutions that scale to foundation-model workloads.
- Act as the connection between research scientists and the infrastructure team.
Requirements
- At least two years of relevant industry experience.
- Highly fluent in and enthusiastic about PyTorch and JAX.
- Experience thinking in asynchronous primitives.
- Strong opinions about clean, minimal, and consistent library design.
- A documented track record of clear, well-documented code, such as through GitHub artifacts.
- Broad machine learning knowledge and an understanding of scalable, reliable ML systems.
- Preferred experience with equivariant architectures, geometric deep learning, graph neural networks, or ML-assisted drug discovery.
- Preferred experience with NequIP, MACE, SchNet, PaiNN, or similar systems.
- Preferred experience building with declarative workflow orchestration frameworks such as Flyte or Dagster.
- Comfort interacting with quantum chemical scientists and their data pipelines is preferred.
Benefits
- Hybrid work arrangement in San Francisco or, for highly skilled candidates, New York City with travel to San Francisco as needed.
- Travel to conferences and corporate on-site activities is part of the role.
Tech Stack
Categories
About Achira
Achira builds atomistic foundation simulation models that simulate molecular interactions to support drug discovery and preclinical research for biotech and pharmaceutical teams. The privately held company, founded in 2024 and based in San Francisco, offers its technology as software and research collaborations. Its work combines AI with large-scale physics-based simulation to accelerate target validation and lead optimization.