AI Full-Stack Engineer

Tanagram's mission is to enable developers to work at the speed of thought. To do that, we're building a tool that captures hard-won lessons buried in codebases, code reviews, incident post-mortems, and Slack chats. We turn those lessons into real-time guardrails that flag or fix risky patterns the moment they reach a pull request — and, eventually, at code generation time — so that engineers can ship faster and avoid disaster.

Imagine an ideal staff engineer. They know the entire codebase and how different systems interact. They follow every PR, so they know what changes are being made and how patterns evolve. They've read all the documentation. They keep up-to-date on Slack conversations. Ultimately, they index all that information in their heads, and deliver impact by showing up everywhere and saying the right things at the right time.

We're building that, but at scale for entire teams and companies. Tanagram is an extension of every team's best staff engineer, available anywhere and anytime.

Please apply here.

About This Role

As an AI Full-Stack Engineer, you'll bridge the gap between AI research and application by using the latest research and models to build our product end-to-end. Currently, our product is almost entirely backend; we do not yet have a UI (although we will start building a UI for our product soon). We mostly build in Python, with some Swift for CPU-bound operations.

We’re a small team of generalists and work across multiple domains. We're looking for meticulous, high-agency people who have good judgment around what problems to solve, the skills (or learning ability) to solve it expediently, and an understanding of the appropriate quality bar given the surrounding business context.

You will own experiments and POCs focused on combining the latest research findings with specific high value problems that our customers encounter each and every day.

We will generally work in-person in San Francisco (our office is in Mission Bay), but are open to remote for the right candidate.

What You'll Do

• Talk directly to users: understand their requirements, ask for their feedback, follow up as needed, and iterate based on what they say.
• Evaluate and build with the best tools in an AI agentic stack: LLM-Evals, Guardrails for AI, CodeAct (agents writing code), memory for agents (like Mem0).
• Build services to ingest data from places like Github, Slack, and Confluence.
• Deploy & use reasoning models like Qwen2.5-7B-Instruct.
• Eagerly identify and implement improvements to our core foundations (e.g. clearing performance bottlenecks for scalability).
• Ensure that systems are efficient, maintainable, and well-monitored.
• Shape our product roadmap by influencing the sequencing of what we want to build, and/or by talking to potential users and proposing new projects.
• Work closely with founders and design to help come up with high-level product concepts (bonus points: and detailed Figma designs) for how interactions will work.
• Other things that may come up from time to time to move our company forward (or keep the lights on).
  
  

What We Offer

• Challenging work with the latest advances in LLM techniques.
• Top-of-market compensation (and a long runway).
• Employee-friendly equity terms (low FMV, early exercise, extended exercise).
• Your choice of Macbook Pro + computer/office equipment stipend.
• Food stipend/reimbursements on meals.
• Health, dental, and vision insurance.
• Unlimited PTO.
• A relatively un-chaotic working environment (we aren't pivoting every week).
• An opportunity to lead and define our company.
  
  

Every company only goes through its founding moments once — the liminal period where foundations get built, rituals get defined, and lifelong friendships are formed. The vast majority of companies out there are past that stage, and you merely join them. At Tanagram, you get to build them.

Qualifications

• At least a few years of IC experience, balancing moving fast with scalable implementations. This role generally maps to a "senior" engineer level, although we're somewhat flexible (and will adjust compensation accordingly).
• Hands-on experience building with LLMs, with a strong intuition for what models can do and how to get the best results out of them.
• Well-crafted (i.e. generally reliable; tastefully designed) work examples.
• Self-direction and motivation: you repeatedly, independently seek out the most valuable thing you could be doing, and start working on it. You do so even when requirements and priorities may be changing rapidly.
• Decisive, clear communication on both technical and strategic matters: at this stage, you'll be helping build the foundations of our company. You should have ideas/opinions about almost everything we do, and you should be forthcoming with your thoughts (even if it means being disagreeable).
• Bonus points:
• Experience working on parsers and compilers (e.g. LLVM; Typescript), type systems (e.g. Sorbet), or VSCode or IntelliJ itself.
• Experience with search and/or indexing technologies.
• If you've previously worked at a startup, or founded one yourself.
  

Compensation

Depending on the relevance and amount of your experience:

• Salary for this position ranges from $160,000 to $240,000 USD.
• Equity ranges from 0.5% to 1.5%. If we move forward with an offer, you will have a choice between more cash or more equity.
  
  

↣ If you've made it this far, please apply here! ↢

Tanagram's mission is to enable developers to work at the speed of thought. To do that, we're building a tool that captures hard-won lessons buried in codebases, code reviews, incident post-mortems, and Slack chats. We turn those lessons into real-time guardrails that flag or fix risky patterns the moment they reach a pull request — and, eventually, at code generation time — so that engineers can ship faster and avoid disaster.