About the role

⚠️ Please read first

This is a full-time, in-person role based in San Francisco (Presidio) - we work from the office 5 days a week.
You must be based in the Bay Area or willing to relocate before starting.
We require US work authorisation , but are open to O-1 visa sponsorship for truly exceptional candidates.

What You’ll Do

As an Applied AI Engineer , you’ll be responsible for building, refining, and scaling the agent systems inside Wordware — from architecture to evals to deployment.

This isn’t a research role. We care about what works in production: fast response times, predictable behavior, traceability, and uptime.

You’ll work across the stack — with infra, frontend, and product — to make sure the agents users build inside Wordware are robust, useful, and usable.

A few examples of what you might work on:

Implement multi-step, tool-using agents that hit real APIs and handle retries, auth, timeouts, and edge cases.
Build RAG pipelines that support grounded answers from structured and unstructured sources.
Design agent memory systems that persist relevant state across runs — e.g. scratchpads, summary buffers, embedding stores.
Add determinism + replay to agents so users can trace and debug behaviors step by step.
Own and evolve our eval framework — both automated checks and human-in-the-loop scoring.
${your ideas}.

Who You Are

Minimum

3+ years of engineering experience , including time shipping production software.
You've built and deployed agent-like systems — multi-step LLM pipelines, tool-using bots, scripted assistants, or similar.
Hands-on experience with:
You write production-grade code and can work across systems without needing a spec.
You thrive in fast-paced, product-first environments where the goal is shipping.

Bonus (not required)

Experience with frameworks like LangChain , CrewAI , or DSPy — or strong opinions about why you don’t use them.
Shipped agents that are live in the wild — used by customers, not just internal demos.
Familiarity with LLM ops , tracing, observability, and failure handling.
You’ve been a founder or early engineer and care deeply about product quality.

The Process

We keep our process simple. Exceptional candidates go from first touch to offer within 2 weeks.

Application
15-min intro call
45-minute technical interview
System design interview
Final conversation
Work trial

About Wordware

AI Agents have arrived. They are powerful enough to read, plan, and act across your everyday life and work. The question is no longer whether they can run canned demos to check the weather, it’s whether real teams can rely on them for critical work. We optimize for adoption over spectacle: useful, accountable systems that help top professional complete real tasks.

About Wordware. Wordware is backed by Spark Capital, Felicis, and Y Combinator, with a $30M seed round. Our beautiful, historic San Francisco office sits in the shadow of the Golden Gate Bridge, it’s full of plants and radiates innovation and calm. We genuinely love working together. At lunch we talk about alignment and AI consciousness, we have surfboards in the office ready to use, we take walking meetings on the beach and hang out after work at the bar or climbing wall. We work really hard because we care deeply about what we do.

About Sauna. Sauna is Wordware’s AI workspace for professionals. Think ‘Cursor for Knowledge Work’ and you’re halfway there. With explicit permission, it connects to tools like Gmail, Calendar, Slack, and thousands more, to gather the right context to move your work forward. It learns who you are, drafts and refines your documents, presentations, and datasets, researches across the web and your org’s internal knowledge, summarizes long threads to cut through the noise, proposes next steps, even acts on your behalf when enabled. You can work with Sauna live, side-by-side, or schedule jobs to run automatically in the background, with clear permissions and easy off-switches. Sauna brings the power of AI Agents to all professionals.

Startups are intense, but the intensity is intentional here. We work in focused sessions, cool down, and go again. You’ll have real ownership, clear outcomes, and the satisfaction of seeing your work land with customers, not months from now, but this week. If you’re joining Wordware, you’re choosing impact over spectacle. You’ll shape the product, the standards, and the way we work together. It’s demanding but humane: we write things down, respect time off, and prefer steady progress to late-night heroics.

Who thrives here? People who like small, correct moves that compound. People who stay calm when the plan changes and make the next good decision without drama. People who treat constraints as design inputs, not roadblocks. If you’ve built something from scratch, carried it through the unglamorous parts, and kept your standards high, you’ll feel at home.

We’re not optimizing for résumé lines; we’re optimizing for judgment. You don’t need to be loud. You do need to be reliable, curious, and willing to own outcomes.

We judge by what you build, not when you build it. Weekends are for life. Burnout doesn’t make good software. We’re aiming for something durable and generational, and that requires knowing when to push and when to recharge. If that balance and this mission resonate, we should talk.

Don't see your role? If what we’re building clicks for you and you’re exceptional, tell us. We’ll consider shaping a role around the right person.

How to stand out? Show us signal. Build something with Sauna, send a short note on what worked and what didn’t, or point out a concrete way to improve the product. Clarity beats volume. Initiative beats polish.

If you’re truly excellent, a LinkedIn link is enough—ping it to lucy[at]wordware.ai