The best ML engineers aren't scrolling job boards — they're shipping models to production. Underdog.io is a closed, invite-only network where the best AI-native startups are introduced directly to vetted ML engineers. One profile. No applications. Hiring managers come to you — with salary ranges upfront.
Why this network exists
Anyone can train a model on a clean dataset. Shipping it to production — dealing with feature pipeline failures, data drift, model degradation at 3am, and inference latency under real load — is a different skill set entirely. Underdog is built for engineers who've done the hard part. The startups in our network know the difference too, and they're specifically looking for you.
Underdog isn't a job board. It's a closed network — the startups with serious ML problems reach out to you, and you decide which ones are worth your time. No spam. No recruiters who haven't read your profile. No noise.
ML roles in the network
Tell us your specialization, stack, and what kind of problems you want to work on. We match you to startups where your specific ML background is the thing they actually need.
RAG systems, agent orchestration, fine-tuning, prompt engineering, LLM evaluation frameworks. Engineers who've shipped LLM-powered features to production users — not just prototyped in a notebook.
Production model training and deployment, feature engineering, A/B experimentation, recommendation systems, search ranking. PyTorch, TensorFlow, Scikit-learn. Bridging research and product.
Training pipelines, model serving, feature stores, evaluation frameworks, vector databases, observability. The engineers who make sure the ML actually works in production at scale.
Deep technical specialization in language understanding, text generation, image recognition, or multimodal systems. The highest-compensated ML specializations — and hardest to find through job boards.
Closing the gap between ML research and product. Strong math background, novel model architecture experience, ability to translate paper ideas into production systems. PhD or equivalent preferred.
Employee #1–3 on the ML team at a seed or pre-seed AI-native startup. You design the ML architecture, pick the stack, and define how the company thinks about AI. High equity, total ownership.
How it works
No listings to scroll. No take-home projects before you even know the salary. No recruiters who can't tell PyTorch from TensorFlow. Here's what actually happens when an ML engineer joins Underdog.
Tell us your ML specialization, production stack, what you've shipped, and what you're looking for next. Takes 60 seconds. No resume upload, no take-home project, no cover letter.
We review every profile by hand. Only the top 5% are accepted — specifically engineers with real production ML experience. When a startup is introduced to you, the conversation already starts with trust.
Every Monday, hiring managers at vetted AI-native startups contact you directly — salary ranges included. You decide who gets your time. 85% of accepted engineers hear from a company in week one.
2026 salary guide
ML engineer salaries rose 5–9% at startups in H1 2026 alone, driven by demand outpacing supply of engineers with genuine production experience. Every Underdog intro includes the salary range before you engage.
Sources: Wellfound, Glassdoor, Carta State of Startup Compensation H1 2025, SalaryCube 2025.
Engineers with real production ML deployment experience — not just model training — command a meaningful salary premium. Startups paying for production experience know it's the scarce skill that actually matters.
Founding ML engineers at AI-native startups often receive 0.25–1.0% equity at seed. The AI infrastructure category in particular is producing significant outcomes — the equity math is real for early joiners.
Base salary ranges at venture-backed startups. Equity and total compensation are additive. All Underdog interview requests include the actual salary range.
Companies in the network
Every company has been reviewed and approved. No staffing firms. No agencies. Just real teams building real AI products — and hiring ML engineers to help them ship.
Not all companies are actively hiring ML engineers at all times. We match you based on your specialization and what's active when you join.
Every time I use Underdog.io I remember that job searching doesn't have to be terrible. Thanks for the product.
Zach B. — Senior Software Engineer, NYC
Common questions
Ready when you are
The AI-native startups with the hardest ML problems hire through Underdog. One profile gets you introduced to all of them — no applications, no take-homes before salary, no recruiters who can't read your stack.
Get in the network — it's free →