Machine Learning Engineer

• Develop and maintain production Python services, pipelines, and CLI tooling.
• Build and maintain Helm charts, container images, and deployment infrastructure as part of daily workflow.
• Deploy and operate workloads on Kubernetes across multiple AWS regions and accounts.
• Instrument pipelines and models so that performance, cost, and benchmark results are measurable.
• Work directly with scientists and engineering leads to decide what to build, and support adoption through documentation and hands-on help.

• Bachelor’s degree with 7-10 years of relevant experience, or a Master’s degree with 5 or more years of relevant experience.
• Strong Python development skills with experience building production-grade tooling or pipelines.
• Working knowledge of Kubernetes, Helm, and container-based deployments.
• Experience developing, deploying, or optimizing machine learning pipelines in a production environment.
• Familiarity with infrastructure-as-code and CI/CD practices.
• Experience with AWS, including multi-region or multi-account deployments.
• Experience with data warehousing and ELT patterns, Snowflake preferred.
• Experience on a small team (roughly 3-10 engineers) where individuals own systems end to end, set their own priorities within a goal, and ship without a dedicated QA, SRE, or release function.
• Routine use of agentic coding harnesses - Claude Code, OpenCode, Codex, or equivalent - as part of day-to-day development, with a view on where they help and where they do not.
• Exposure to Rust is a plus.
• Experience building LLM agents, tool definitions, or evaluation harnesses is a plus.
• Willingness to learn computational biology and protein science domains.
• Effective communicator who can work independently in a remote environment and collaborate with both engineers and scientists.

Spectrix offers flexible schedules, competitive salaries, and a committed work environment to its employees; the company provides automatic benefits of life insurance and short and long term disability insurance. Optional benefits for full time employees include health, dental, and vision insurance, college loan assistance, and an employer matched retirement savings plan. Spectrix is not a placement agency.

Spectrix is participating in the E-Verify program of the U.S. Department of Homeland Security (phone number: 888-897-7781, and website: www.dhs.gov/E-Verify).

Pay: $48.00 - $58.00 per hour

Expected hours: 40.0 per week

Spectrix Analytical Services is seeking a ML Platform / Agentic AI Engineer. This contract role focuses on building and operating the compute platform, MLOps, and agentic tooling that support protein therapeutic design. This is a full-time (40 hours/week) remote position open to candidates in the United States or Canada. Working hours within one to two hours of US Eastern time are preferred; West Coast candidates who can overlap with Eastern business hours will also be considered.

The contractor will work across a Kubernetes-based workflow orchestration platform, a scientific tools monorepo of containerized computational biology applications, and the data and model infrastructure that connects them to downstream analytics in Snowflake. Depending on background, work will center on one or more of the areas below; we expect contractors to work across area boundaries as priorities shift.

Areas of work:

• Platform and workflow engineering. Extend the Kubernetes orchestration platform and the workflows that run on it, including migrating existing application backends onto it. Make the platform region-agnostic so compute can be scheduled in new AWS regions and accounts with configuration rather than code changes.
• MLOps and model serving. Build the registration, versioning, and serving path for models so that model outputs are available to scientific applications and to downstream reporting. Integrate model output with the data layer rather than leaving it in per-project artifacts.
• Automated benchmarking. Build automated benchmarking and retraining for computational tools, LLM models, and agent skills used in de novo design. This includes GPU-backed benchmarking of larger models, test-set and dataset tracking, and a persistent record of benchmark results over time.
• Agentic AI integration. Incorporate agentic AI into existing scientific applications: tool and skill definitions, evaluation harnesses, and the plumbing that lets agents call internal services safely.
• Data engineering. Build reliable S3-to-Snowflake data pulls and automated registration of de novo design results, with monitoring and backfill behavior that does not require manual intervention.
• Performance optimization. Take existing production pipelines and reduce their runtime and cost - profiling, GPU utilization, batching, scheduling, and container image work.