How to Hire ML Engineers in San Francisco: A 2026 Playbook

San Francisco has more machine learning engineers than any other city in the world, and it is still one of the hardest places to hire them. You are competing with frontier AI labs, MAANG and hundreds of well-funded startups for the same few thousand people, and most of them are not looking for a job.

I have spent years sourcing ML and AI talent for startups and big tech, including at Meta. This is the playbook I use when a founder asks me how to hire ML engineers in the Bay Area without wasting three months.

Why hiring ML engineers in San Francisco is so hard

• Demand is concentrated. AI labs, big tech and startups all recruit from the same small pool of people with real production ML experience.
• The best people are passive. Strong ML engineers get messages from recruiters every week. They ignore generic outreach and rarely apply to job posts.
• Titles are messy. "ML engineer", "AI engineer", "applied scientist" and "research engineer" can mean very different jobs at different companies.
• Interview loops are slow. Many teams lose their top candidate to a faster offer, not a better one.

Step 1. Define the outcome, not the title

Before you write a job description, write down what this person must deliver in their first six to twelve months. Then map it to one of four families of AI work:

1. Research — new methods, model quality, papers.
2. Modeling and applied ML — training, fine-tuning and evaluating models on your data.
3. AI product engineering — building features on top of models: LLM apps, agents, retrieval.
4. ML infrastructure — training pipelines, inference, GPUs, MLOps.

Most failed searches start with a title that mixes two or three of these. If you are not sure which one you need, read my guide AI and ML Roles Explained.

Step 2. Write a job description engineers actually read

ML engineers skim job posts for three things: the problem, the data and the compute. Lead with them.

• What problem the team is solving and why it is hard.
• What data and models they will work with, and what stack is in production today.
• Access to GPUs, how much research freedom they will have, and who they will learn from.
• Level, salary range and equity. California requires employers with 15 or more employees to include a pay range in job postings, and candidates expect it anyway.

Cut the list of twenty "requirements". Keep the three or four that really matter.

Step 3. Source far beyond LinkedIn

If you only search LinkedIn, you see the same profiles every other recruiter sees. The engineers you want often leave better signals elsewhere:

• GitHub — contributors to the libraries and models you actually use.
• arXiv and Google Scholar — authors of applied papers close to your problem.
• Conference speaker and author lists — NeurIPS, ICML, ICLR, CVPR, ACL and their workshops.
• Kaggle, ICPC and olympiad rankings — a strong signal for algorithmic and modeling skill.
• Open-source and research communities — Hugging Face, Discord servers and local Bay Area meetups.

AI sourcing tools like Noon.ai and Juicebox help you search these sources at scale, but a recruiter still needs to judge who is a real fit.

Step 4. Write outreach that gets replies

Passive candidates reply when a message is clearly about them. A good first message is short and specific:

• Mention something real they built or wrote: a repo, a paper, a talk.
• Explain the problem in one sentence and why their experience fits it.
• Share the level and compensation range up front.
• Ask for a 15-minute call, not a full application.

Step 5. Screen for real work, not keywords

A resume full of the right keywords tells you very little. Before the first technical interview, collect evidence of real work: GitHub, publications, Kaggle results, talks and past projects. A recruiter screening call with AI notes, such as Metaview, keeps this context in one place for the hiring team.

Then keep the technical loop focused and fast:

1. A deep dive into a past project the candidate owned end to end.
2. One coding or modeling exercise relevant to your work.
3. An ML system design interview for senior roles.
4. A conversation with the founder or hiring manager about the mission and growth.

Try to complete the whole loop within two weeks. Strong candidates are usually in several processes at once.

Step 6. Win the offer

In the Bay Area, compensation is necessary but rarely sufficient. Benchmark salary and equity against current market data, such as Levels.fyi and your investors' compensation surveys, so you are not out of range. Then compete on what money cannot easily buy:

• Interesting problems and ownership of real models in production.
• Access to compute and strong teammates.
• A clear growth path and a fast, respectful hiring process.

Make the offer within a few days of the final interview and explain the equity clearly. Uncertainty is the most common reason I see ML candidates walk away at the last step.

Common mistakes to avoid

• Hiring a "unicorn" who should do research, infrastructure and product at the same time.
• Relying only on inbound applications and LinkedIn searches.
• Long take-home assignments that strong candidates simply skip.
• Hiding the salary range until the offer stage.
• Letting the process stall for a week between interviews.

FAQ

How long does it take to hire an ML engineer in San Francisco?
It depends on seniority and how specialized the role is. The biggest delays usually come from an unclear role definition and slow interview loops, not from a lack of candidates.

Should I hire an ML engineer or an AI engineer?
If you need to train or fine-tune models on your own data, hire an ML engineer. If you are building product features on top of existing models and APIs, an AI engineer is usually the better fit.

Where can I find ML engineers who are not on LinkedIn?
Look at GitHub contributors, arXiv authors, conference speaker lists, Kaggle and ICPC rankings, and open-source communities like Hugging Face.

Should I use a recruiting agency to hire ML engineers?
An agency makes sense when the role is senior or niche, when you need to move fast, or when your team doesn't have time to source passive candidates. Choose one that specializes in AI and ML and can show you the evidence behind each candidate.

The takeaway

Hiring ML engineers in San Francisco is a competition for attention. Define the outcome, go where the talent actually is, show candidates their real work matters, and move faster than everyone else. If you need help, Relo finds ML engineers, AI researchers and robotics talent for Silicon Valley startups and AI labs.