AI Engineer

Stealth Mode • Full-time • San Francisco Bay Area, US • 1d ago

About Us

Built by a team with deep roots in human-computer interaction — from Apple Vision Pro and Pokémon GO to large-scale consumer interactive products — we are transforming personal AI from a chatbot into continuous, wearable intelligence that understands and improves your real life.

If large language models are the engines that grasp the sum of world knowledge, but there isn't an equivalent grasp of you: your daily activities, energy, behavior patterns, goals and the moments when better guidance would actually change what you do. We are building technology that answers the ancient Delphic maxim: Know thyself. Our founding team has built these systems from early demos through SDKs to shipped engines, we are synthesizing long-term personal context with universal knowledge to help people improve themself by enabling them to see themselves more clearly, make better decisions, and become healthier and more capable in everyday life——and eventually, the best versions of themselves.

We are tackling hardware and software paradigms the field hasn’t figured out yet—multimodal reasoning as a collaborative effort between edge and cloud, learning from a person’s day in real-time without burning the battery or the budget. Nobody has shipped this yet. That’s why we’re doing it.

What You'll Do

You will work directly with the founding team. There is no tech lead between you and the work. You are the team, and you’ll grow at the rate you ship. Your job is to answer open questions, fast, by building.

End-to-end experimentation: You drive the loop — generative models, context engineering, the perception layer underneath — from "I wonder if" to a fast, honest answer. We don’t expect polished solutions. Tight loops that teach us something every week.

Ship Novel Hardware/Software Paradigms: Tackle shifting weekly challenges. How many tokens does this burn per user per day, and can we cut it in half? Does more historical context actually improve what the model says back? What latency can we hold under, and what does the user feel when we miss it?

You May Be A Good Fit If...

You have a bias for action (Low Ego): You hear a vague idea and your hands start moving. You aren't asking for requirements or scoping the work. You'd rather ship something ugly by Friday than something clean next month. (If you need concrete tasks to be productive, this isn't it).

You turn vague into code (Genius): Someone says "I wonder if X would work," and an hour later there's a script that tells you whether it would. Defining the problem is part of what you do, and you do it by building.

You take absolute ownership: Nobody tells you what to do next; you figure out what matters most. When something breaks, it's yours to fix. When something works, it's yours to extend.

You are a force multiplier: You work fluently with AI coding tools (Cursor, Claude Code, Codex) and have set up your own workflow. Crucially, you possess the judgment to know when not to use them. We hire for judgment, not just throughput.

You have hands-on experience with Generative Models: You've built something real with LLMs or VLMs, run open-source models (DeepSeek, Qwen, etc.), and stood up pipelines yourself.

You understand Context Engineering: You've wrestled with RAG, memory systems, MCP servers, or making a model reason over things it didn't see in pretraining.

You have a baseline of technical fluency: 1–2 years of professional or serious internship experience (fresh grads are welcome with a substantial portfolio of shipped work, not just coursework), Python fluency, and fluent English.

(Bonus) You have experience in computer vision, Docker, or at least one major cloud platform (AWS, GCP, or Azure).

How We Are Different

We believe the highest-impact AI will be deeply personal, not just a massive cloud oracle. We work as a tightly knit unit to bridge the gap between world knowledge and an individual's real-time context.

Because we are tackling unsolved hardware and software paradigms, there is no playbook. We are looking for a fearless builder to run ahead and test the boundaries of what’s possible. We expect you to be bold enough to pitch wild, creative solutions, and fast enough to write the code that proves them right or wrong.

Mistakes aren't just tolerated; they are expected, because they prove we are moving fast enough. We will debate the right problems to solve as a tightly knit team, but your mandate is pure execution: push the limits, try things that might break, and let what we learn by building be our ultimate source of truth.

Come Work With Us

We know that to solve complex human challenges, we must first ensure the well-being and peace of mind of our own team. We work as a single, tightly knit unit on one massive objective.