San Francisco, CAremote

Salary
Posted
Jul 26, 2026
Location
San Francisco, CA
Last confirmed open
Jul 28, 2026

What this job asks for AI summary

This role owns the end-to-end model development pipeline for a production automotive engagement at an MIT-spinout AI company. Day-to-day work spans translating partner feature specs into training requirements, building and maintaining large-scale data pipelines, running fine-tuning and evaluation cycles for an on-device audio-to-function-calling model, and delivering polished partner-facing analyses and dashboards. It suits a hands-on ML practitioner comfortable with embedded/on-device constraints and direct client communication.

Mid level · 2+ years · Full-time

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

Must have (4)
machine learning · 2+ yrsmodel trainingdata pipelineson-device ML
Nice to have (2)
LLMsaudio/speech models

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

What won't set you apart
machine learning80%data pipelines40%

Most people in this occupation already list these. Still required — just not what gets you shortlisted.

What the occupation pays Median $122,874 (middle half $87,544–$162,374).

Estimated from BLS employment for this occupation and area, per-skill prevalence across our listing corpus, and published wage benchmarks — as of Jul 28, 2026. It is a model, not a headcount.

Why we read it this way (6)

No work location or metro is specified in the posting. Liquid AI is headquartered in Cambridge, MA (MIT CSAIL spinout), but the JD does not explicitly state a location or remote policy — location fields left blank.

The role blends ML research/modeling (15-2051) with substantial software engineering for data pipelines and tooling (15-1252). Primary day-to-day work is model training and evaluation, which tips toward 15-2051, but the pipeline-engineering emphasis makes 15-1252 a genuine runner-up.

The '2+ years' figure is stated as a guideline with explicit openness to exceptional early-career candidates; the overall years minimum is set to 2 accordingly. Seniority is assessed as Mid given the 2+ year floor and the scope of autonomous ownership expected within one year.

The must-have skills are drawn from the explicit 'Must-have' section. Nice-to-have skills (audio/speech, function calling fine-tuning, multilingual, automotive/AV/defense lab background) are marked preferred per the 'Nice-to-have' heading. 'Model training' and 'data pipelines' are named as concrete activities in the must-have block; 'on-device ML' maps to the stated requirement for automotive/embedded/on-device ML context.

The posting explicitly states that 'building applications around model APIs does not qualify' as ML experience — a notable screening signal for recruiters.

Ignored 2 non-technology phrase(s) as skills (responsibilities/concepts, not named tools): function calling / tool-use fine-tuning, multilingual model development.

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 ↗