Boston, MAremote

Salary
—
Posted
Jul 31, 2026
Location
Boston, MA
Last confirmed open
Sep 24, 2026

What this job asks for AI summary

An ML infrastructure engineer role at an early-stage AI/physics research startup, responsible for building and owning the distributed training and model-serving systems that research teams depend on. The work spans GPU scheduling, inference serving (vLLM, SGLang), developer tooling for researchers, and production reliability — with a strong expectation of hands-on coding and on-call ownership. Suited to engineers with production ML infrastructure experience at meaningful scale.

Senior level · 3+ years · Boston-Cambridge-Newton, MA-NH · Full-time

Quick apply — this platform usually takes a CV and a few fields.

Must have (4)
PyTorch or Raydistributed trainingvLLM, Sglang or TritonGPU scheduling
Nice to have (3)
CUDAGCP or AWSTerraform

“or” means any one of them counts — you don't need all of them.

We read this from the posting text with AI. Skim the description below before ruling yourself out.

How this req sits in the market our data

Roughly 70 people in the Boston-Cambridge-Newton, MA-NH area plausibly meet what this posting asks for (software developers). range 20–120

Applicant volume Moderate — A normal amount of company. The rare requirements below are what will separate a shortlisted application from the rest.

What gives you an edge
vLLM3%PyTorch9%

Rare in this occupation — lead with these, and say what you built with them.

What the occupation pays Median $169,742 (middle half $134,780–$187,781).

Estimated from BLS employment for this occupation and area, per-skill prevalence across our listing corpus, and published wage benchmarks — as of Aug 2, 2026. It is a model, not a headcount.

Why we read it this way (7)

The role title 'Member of Technical Staff, ML Engineer' carries no explicit seniority level, so advertised seniority is Unspecified; however, the 3+ years of production ML infrastructure experience and full end-to-end ownership expectations support a Senior classification.

The alt SOC (15-1244) was considered because the role involves significant infrastructure operation, but the emphasis on writing production code, building services and abstractions, and owning the full stack from spec to ship points more clearly to Software Developers (15-1252).

Remote work is listed as considered 'on a case-by-case basis' — not a firm remote offering — so remote is set to false.

No compensation figures are provided beyond a general reference to competitive salary, benefits, and equity.

Distributed training is listed as a required capability; PyTorch and Ray are named as acceptable frameworks — PyTorch is emitted as the primary with Ray as an alternative. Similarly, vLLM is primary for model-serving with SGLang and Triton as alternatives.

CUDA, GCP/AWS, and Terraform appear only under the 'Nice to Have' section and are marked preferred accordingly.

'Training-as-a-service APIs', 'inference gateways', and 'job schedulers' are mentioned under Nice to Have but name no specific named technology, so they are omitted per the no-generic-concepts rule.

Read the full posting

The employer publishes the full description on their own site — read it there ↗. Or sign in to read it here — it's free, and it also lets you track this application.

Apply

Apply on employer site ↗