About Tanagram:

Tanagrams mission is to enable developers to work at the speed of thought. To do that were building a tool that captures hardwon lessons buried in codebases code reviews incident postmortems and Slack chats. We turn those lessons into realtime 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. Theyve read all the documentation. They keep uptodate 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.

Were building that but at scale for entire teams and companies. Tanagram is an extension of every teams best staff engineer available anywhere and anytime.

About This Role:

As an AI Engineer Intern youll bridge the gap between AI research and application by using the latest research and models to build our product endtoend. 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 CPUbound operations.

Were a small team of generalists and work across multiple domains. Were looking for meticulous highagency people who have the skills (or learning ability) to solve problems expediently and an understanding of the appropriate quality bar given the surrounding business context.

We will generally work inperson in San Francisco (our office is in Mission Bay).

What You Might 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: LLMEvals 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 Qwen2.57BInstruct.

• Eagerly identify and implement improvements to our core foundations (e.g. clearing performance bottlenecks for scalability).

• Ensure that systems are efficient maintainable and wellmonitored.

What We Offer:

• Challenging work with the latest advances in LLM techniques.

• Topofmarket compensation.

• Computer/office equipment stipend.

• Food stipend/reimbursements on meals.

• Health dental and vision insurance.

• Unlimited PTO.

• A relatively unchaotic working environment (we arent pivoting every week).

• An opportunity to help define our company.

Qualifications:

• Handson experience building with LLMs with a strong intuition for what models can do and how to get the best results out of them.

• Wellcrafted (i.e. generally reliable; tastefully designed) work examples.

• Selfdirection 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 youll 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 youve previously worked at a startup or founded one yourself.

Compensation:

Depending on the relevance and amount of your experience the salary for this position ranges from $2000 to $2500 per week.

Required Experience:

Intern