What are the responsibilities and job description for the GPU Infrastructure & Architecture Engineer position at Covenant Biosciences?
About Covenant Biosciences
Covenant Biosciences is an AI-native drug discovery company targeting the nearly 80% of human proteins that remain beyond the reach of current therapies and more than 7,000 rare diseases without approved treatments. We build specialized AI platforms for small molecule design (CIPHER), precision biologics (PRAXIS), and next-generation antibodies (AEGIS), each trained on tens of billions of molecular and clinical data points. Our team operates with the urgency of a startup and the rigor of regulated medicine. We exist for the patients still waiting.
The Role
You will own the GPU computing backbone of our drug discovery platform. This is a hands-on engineering role responsible for designing, scaling, and operating the distributed GPU infrastructure that powers our deep learning models and molecular simulations. You will work directly alongside ML scientists and computational chemists, translating research requirements into reliable, high-performance systems.
This is an on-site role based in our Philadelphia as well as our upcoming data center environment. It includes participation in scheduled maintenance windows and on-call rotations to support mission-critical operations.
What You Will Do
- Architect and scale GPU clusters supporting large-scale generative AI model training, graph neural network workloads, and molecular dynamics simulations.
- Migrate and re-engineer existing ML pipelines to be GPU-native and distributed across multi-GPU and multi-node configurations.
- Optimize CUDA kernels, memory management, and throughput for scientific computing workloads including virtual screening and MD simulations.
- Benchmark and select hardware and cloud configurations across AWS EC2, GCP, and on-premises NVIDIA A100/H100 clusters.
- Configure and maintain high-performance networking and low-latency storage solutions suited to AI workloads.
- Administer Linux-based servers and automate infrastructure management using tools such as Ansible and Terraform.
- Implement and maintain security controls including firewalls, VPNs, access control, and encryption in compliance with relevant standards.
- Monitor system and GPU performance, troubleshoot hardware, OS, and networking issues across complex distributed systems.
- Perform capacity planning and document infrastructure standards, procedures, and best practices.
- Collaborate with ML, computational biology, and data engineering teams to identify bottlenecks and co-design architecture as research needs evolve.
What We Are Looking For
- 3 or more years of hands-on experience with GPU computing, CUDA programming, or distributed deep learning infrastructure.
- Proficiency in PyTorch, JAX, or TensorFlow with demonstrated experience in multi-GPU orchestration using DDP, FSDP, or Horovod.
- Experience with cluster management tools such as Slurm, Kubernetes, or Ray.
- Strong Linux system administration skills and comfort with containerization via Docker or Kubernetes.
- Networking expertise covering design, routing and switching, and performance optimization in high-throughput environments.
- Familiarity with the NVIDIA ecosystem including CUDA, drivers, and NCCL.
- Experience with scientific or research computing workloads is a strong advantage.
- Bachelor's degree or higher in Computer Science, Electrical Engineering, Information Technology, or a related field, or equivalent practical experience.
Compensation
Competitive base salary commensurate with experience, and full benefits.
Why Join Us
This role puts your infrastructure work directly in the path of drug candidates that would not exist without it. You will have work on genuinely hard problems, and help build the foundation of a company targeting diseases that the rest of the industry has walked away from.