Applied AI Engineer — Shape the Self-Improving Design Agent

Mixpeek; San Francisco (CA); Hybrid; USD 150,000 - 230,000; Full time

Benefits offered by this job

Health coverage

Dental coverage

Vision coverage

20 days PTO

Team offsites

Job summary

Magic Patterns is seeking an AI engineer in San Francisco to own and advance the evaluation system powering the self-improving agent loop. You will push the boundaries of AI in design and collaborate with product teams to elevate agent quality from a user standpoint.

The role emphasizes building eval mechanisms, context-aware design enhancements, and contributing to post-training on design output, with a focus on real user impact and rapid iteration.

Qualifications

• Experience building eval systems and self-improving agent loops.
• Experience with AI models and design outputs.
• Strong first-principles thinker who distills complex problems.

Responsibilities

• Own our eval system that feeds into our self-improving agent loop.
• Push and define the limits of what AI can do in design.
• Collaborate with the product team to improve agent quality from the user perspective.
• Decide what to work on to improve our AI Design agent. The listed items are open problems, not assignments.

Skills

AI design

Eval systems

Context engineering

First-principles thinker

Tools

PyTorch

TensorFlow

Job description

Hey, I’m Teddy, CTO & co-founder at Magic Patterns! Our engineering team has grown a lot over the last few months here in San Francisco.

About Magic Patterns
Our mission is to let anyone build delightful software. We're a lean team where everyone contributes directly to the product.

If you’re an AI engineer, our roadmap is probably a goldmine of things you’ve always wanted to work on:

• An AI agent that thousands of users use religiously everyday

• Evals with a self-improving agent loop

• Context engineering

• Post-training on design output

The most common piece of feedback we get is "I tried a bunch of tools but your design quality output is by far the best." We want to push the limits even further.

About this role

• Own our eval system that feeds into our self-improving agent loop

• Push and define the limits of what AI can do in design. Today, we are state of the art, and we want to continue to push that.

• Collaborate with the product team to bring novel solutions to improve agent quality from the perspective of our users

• Decide what to work on to improve our AI Design agent. The list above are open problems, not assignments

This role might be a fit if you...

• Use AI tools in your daily workflow

• Have experience with AI models and/or have worked on an AI agent before

• Have experience building eval systems, post-training, context engineering

• Strong first-principles thinker that can distill complex problems

Our Values

1. Obsessive — we love exploring every corner of the hardest problems.

2. Growth-minded — where you are today matters less than how fast you are growing

3. Ownership — insist on high standards and craftsmanship.

4. Spread the magic — for many people, Magic Patterns is “magic.” Be proud of your work and have fun!

Compensation and benefits:

• Rapid career growth at one of the fastest growing Series A companies

• Competitive salary and equity package

• 20 days off PTO

• Health, dental, and vision coverage

• Fun team offsites and events — we went to Mexico last year!