SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 200,000 - 300,000 / annual
Capy is building an AI software engineer that handles real work—taking threads from Slack, Linear, web, desktop, or API, making code changes on cloud machines or local devices, opening PRs, and managing CI and review feedback until merge. The company is a small, YC-backed team in San Francisco shipping to production codebases daily, and they dogfood their own product.
As a Member of Technical Staff, you'll own the entire product experience across Capy's full stack. This is a full-stack role spanning interfaces developers use to direct agents, the agent loop and its tools, the backend that re-plans from, the infrastructure that boots machines, and the layer interfacing with frontier models. You'll make product decisions without a PM, design without waiting on mockups, and solve hard engineering problems end-to-end rather than handing them off.
Key responsibilities include:
- Shipping across the entire stack: agent loop tools, React surfaces, migrations, Kubernetes changes, all in the same day.
- Setting the standard for agent UX: designing interfaces for twenty-five agents running in parallel (threads, tasks, diffs, reviews, live machine desktops streamed from Rust remote desktop).
- Designing in code: taking features from rough idea to polished, animated, keyboard-first interfaces in React web and Electron desktop apps.
- Building the backend: Effect services over MySQL (Vitess on PlanetScale), realtime fan-out over NATS, typed APIs, git proxy, worker threads.
- Expanding agent capabilities: new tools, better verification, tighter CI/review loops, smarter task orchestration, provider layer across Anthropic, OpenAI, Gemini.
- Running infrastructure: Kubernetes clusters, IaC, VM images, sandbox platforms.
- Directing Capy on Capy: writing specs, steering agents, reviewing PRs, fixing the harness.
- Talking to users: gathering feedback from developers running Capy on real codebases and shipping solutions quickly.
You should be agent-native (Claude Code, Codex, Cursor, or your own harness is how you already work), fluent in TypeScript 7, Node 24, React, and typed functional code (Effect or similar). You've shipped against MySQL, run something on Kubernetes, and touched message buses like NATS. You're comfortable at every layer—debugging slow queries, broken websockets, failing Helm releases, and misbehaving tool calls in the same afternoon. You have taste in interfaces (noticing four-pixel misalignments, caring about motion, hierarchy, empty states) and can point to shipped work proving it. You make independent product judgments without a PM or designer defining the problem first. You're fast without shortcuts: you verify work, read diffs, and prefer deleting code to adding unused abstractions. You must be in San Francisco, in a room with the team.
Nice-to-haves include: Effect in production or Rust; sandboxed execution, VM images, remote desktops, or git proxy experience; a model-provider layer you've built or direct work against Anthropic, OpenAI, Gemini APIs; design-tool fluency (Figma) on top of code prototyping; public work (open source, posts, demos, active X account); founder or early-employee experience.
REQUIREMENTS:
- Agent-native: proficiency with Claude Code, Codex, Cursor, or equivalent; sharp opinions on their shortcomings.
- TypeScript 7 on Node 24 and React in production; typed functional code (Effect or similar) you can read and write without ramp-up.
- Shipped work against MySQL, Kubernetes experience, familiarity with message buses like NATS.
- Comfortable debugging across all layers: database queries, websockets, infrastructure, agent tool calls.
- Strong interface taste: attention to detail (alignment, motion, hierarchy, empty states) with shipped examples.
- Independent product judgment: ability to define and solve problems without PM or designer input.
- Fast execution with quality: verification, diff review, preference for deletion over premature abstraction.
- Must be based in San Francisco, working in-office with the team.
