Scientific Platform Engineer

The SBGrid Consortium at Harvard Medical School supports a large international research community by curating and distributing a scientific software platform used across structural biology, cryo-EM, and related fields. The platform includes approximately 650 software titles and 6,000 versions across macOS and Linux and is deployed across laptops, workstations, HPC clusters, and cloud environments. We are hiring a Scientific Platform Engineer to help lead the modernization, security, reliability, and engineering evolution of this platform. This is a platform engineering role with substantial independent responsibility for CI pipelines, reproducible packaging, deterministic installation, release engineering, runtime hardening, observability, and software supply-chain integrity. The role is designed to be primarily engineering and platform-development work, not routine support, and it directly impacts software delivery and platform reliability across a globally distributed scientific infrastructure.

What You Will Work On: This is an engineering-heavy role – expect 90%+ project/building time vs break-fix.
Build & Test Automation

- Design and implement CI pipelines for scientific software across macOS and Linux.

- Develop regression and smoke test harnesses for packaged software.

- Catch failures before distribution rather than after client installation.

- Support fast-moving development branches (e.g., nightly builds) safely.

Reproducible Packaging

- Help define and enforce a canonical build contract.

- Improve dependency tracking and version control.

- Enable deterministic rebuilds across environments.

- Contribute to artifact integrity and metadata tracking (e.g., SBOM readiness).

Runtime Platform Hardening

- Add tests and versioning discipline to SBGrid’s runtime wrapper system (“capsules”).

- Introduce feature flags and safer rollout mechanisms.

- Improve logging, observability, and error classification. Internal Tooling & Observability.

- Develop dashboards and structured signals around build failures and common error states.

- Reduce reliance on tribal knowledge by encoding workflows into systems.

Technologies You’ll Use (and can help shape):
Core platforms

- Linux (expert-level): shells, process model, filesystems, toolchains, debugging, perf basics

- macOS (strong): building, testing, and release workflows across Intel + Apple Silicon

Build/release + automation

- CI/CD: GitLab CI (or equivalent CI systems and concepts)

- Scripting and automation: Bash + Python (primary)

- Performance-oriented implementation as needed: Go and/or Rust (selectively, for the hot paths)

Packaging and reproducibility

- Current + future packaging direction: evaluating/adopting Nix/Spack/Homebrew-style approaches

- Dependency management, artifact metadata, caching, provenance, reproducible builds Execution environments

- Containers and virtualization: Docker/Podman, VMs, and orchestration concepts (framework-agnostic; Kubernetes/OpenShift not required)

- (Nice-to-have) Apptainer/Singularity in scientific/HPC contexts

Version control + engineering hygiene
- Git, code review workflows, testing discipline, documentation-first habits

Nice-to-have context

- Research/HPC exposure (Slurm, shared filesystems, scientific software stacks)

- AWS familiarity (useful, not required)

“You Don’t Need to Know Everything”

We expect this job description to match people coming from different directions including Unix/HPC/research computing admins who like building durable automation, build/release engineers who want harder problems than a typical web app pipelines, and developers with systems instincts who are comfortable close to the OS and tooling.

If you’re strong in either the systems/ops side or the programming/build tooling side, and you want to grow into the other half, we want to hear from you.
Basic Qualifications:
- Minimum of five years’ post-secondary education or relevant work experience.

Add