Machine learning engineers at AI-native startups earned a median salary of $250K with a p90 ceiling of $390K in 2026, based on 1,000 active ML engineering roles tracked by Recruiting from Scratch. Compensation has risen sharply at Series B and later stages as AI infrastructure companies compete for a limited pool of engineers who can build production ML systems.
| Company | Stage | Median | p90 Ceiling | Roles Tracked |
|---|---|---|---|---|
| Tenstorrent | Series D | $300K | $500K | 20 |
| OpenAI | Late Stage | $308K | $490K | 225 |
| Anthropic | Series H | $363K | $485K | 92 |
| Thinking Machines Lab | Late Stage | $413K | $475K | 16 |
| xAI | Series E | $310K | $440K | 19 |
| Perplexity | Series E | $313K | $405K | 26 |
| Decagon | Series D | $300K | $400K | 47 |
> Hiring ML Engineers? Recruiting from Scratch tracks ATS boards and H1B filings for 163+ companies in this space. We typically present qualified candidates within 5 business days. Work with us →
| Stage | Median | p90 Ceiling | Roles Tracked |
|---|---|---|---|
| Series H | $630K | $850K | 28 |
| Series F | $272K | $423K | 28 |
| Series D | $275K | $400K | 70 |
| Late Stage | $273K | $400K | 95 |
| Venture-backed | $269K | $370K | 26 |
| Public | $253K | $358K | 215 |
The ML engineering market in 2026 is bifurcated by capability level. Engineers who can ship production ML systems — not just run experiments in notebooks — command a significant premium. Companies hiring at $300K+ are almost exclusively looking for engineers with production experience: real-time inference pipelines, model serving at scale, feature stores, and evaluation systems.
The recruiting mistake we see most often: hiring managers writing ML job descriptions targeting a "research scientist" profile when they actually need an ML engineer who can ship. These are different people with different compensation expectations, and they don't compete for the same roles.
If you're an ML engineer with production systems experience, the market strongly favors you in 2026. The 1,000+ active roles in our dataset represent strong demand — but many of them are competing for a small pool of qualified candidates. If your background is primarily research-focused (academic ML, large-scale model training), you may find that companies hiring "ML engineers" have production requirements that don't match your experience. The most valuable positioning for senior ML engineers right now: demonstrable experience getting models from training to production at scale.
ML engineers at AI-native startups earned a median of $250K with a p90 ceiling of $390K in 2026, based on 1,000 active ML engineering roles tracked by Recruiting from Scratch.
ML engineers with production systems experience typically earn 15–25% more than comparable senior software engineers at the same company. The premium is highest at companies where ML is the core product — not a feature.
ML engineers build and maintain production systems — model serving, training pipelines, feature stores, evaluation infrastructure. Research scientists focus on novel modeling approaches and experimental work. The roles require different skills and attract candidates from different backgrounds, though the best ML engineers have some research depth and vice versa.
Based on our ATS data, demand is concentrated at companies building AI infrastructure and applied AI products at Series B and later. The highest-paying companies tend to have production ML as a core business requirement, not a supporting capability.
Recruiting from Scratch specializes in ML and AI engineering recruiting for AI-native startups and growth-stage technology companies. Get in touch to discuss your search.
Beyond model training, the highest-demand skills are: production inference optimization (CUDA, TensorRT), distributed training (FSDP, DeepSpeed), ML evaluation systems, and MLOps (model monitoring, drift detection, A/B testing). Engineers who can close the loop from research to production are rarer and command the largest premium.
At the p90 level, yes — AI labs (OpenAI, Anthropic, xAI) pay the highest ML engineering comp in our dataset. AI product companies (applying models vs. building foundation models) typically pay 10–30% less at the senior level, but the equity risk profile is different.
Data from Recruiting from Scratch's market intelligence platform: 1,000 active machine learning engineering job postings (title pattern + role_family classification) across our ATS network of 89,000+ companies, supplemented by U.S. DOL H1B LCA compensation filings. Updated June 2026.
RFS places ML Engineers exclusively at VC-backed companies and pre-IPO startups. If you're evaluating an offer or exploring what's out there, we can help. Connect with a recruiter →
Tell us about your open roles and we'll start sourcing within 48 hours.
