Member of Technical Staff

Job Description

Member of Technical Staff @ Hussh

\n

\n

About Hussh

\n

\n

Hussh builds private AI that is owned and controlled by the individual — not the company, not advertisers. Our "One Agent" is a lifelong AI helper that securely manages personal documents and everyday work like email, calendar, and communication, sharing data only when and how the user chooses and also connecting to every other ai agent and work with them on your behalf. Our "One Puppy" is a personal supercomputer that runs these AI helpers at home, keeping powerful AI under your roof and under your control.

\n

\n

We were founded in response to a real case of identity theft, and we're driven by a simple belief: people should always know who accesses their information — and make that decision themselves. Our mission is to make AI useful at scale while preserving ownership, privacy, and autonomy. It started off as Project Hushh at Google as a 20% project with a mission to make information an asset for everyone starting with consumers.

\n

\n

The role

\n

\n

Full-time, on-site in Kirkland, WA. You will design, build, and optimize the systems behind private, user-owned AI agents and the Puppy device — from ambiguous problem to dependable product in the hands of real people.

\n

\n

This is a founding-engineer seat. You own the outcome: product interfaces, backend services, infrastructure, agent systems, local models, and the last mile — migrations, security, deployment, observability, rollback, real-user verification. A working demo is the beginning, not the end.

\n

\n

What we're looking for

\n

\n

- Hands-on depth across Python, TypeScript, React, PostgreSQL, Docker, and GCP. You move comfortably between frontend, backend, and infrastructure.

\n

- Real experience with agent systems — MCP, A2A, and modern agentic engineering. You understand tool contracts, scoped authority, and verifiable execution.

\n

- System design instincts: async parallel workflows with explicit state, cancellation, idempotency, checkpoints, and recoverable failures.

\n

- You've run and integrated local models, and you diagnose constraints with measurements, not vibes.

\n

- Public proof: meaningful open-source work with live deployments. Be ready to walk us through the code, the architecture, the tradeoffs, and what broke in production.

\n

- Human taste: you notice friction, simplify complex workflows, and care about interaction details and failure states that people can understand.

\n

- You ship without step-by-step direction, communicate risks early, and measure your impact in verified task completion, latency and cost per task, and release reliability.

\n

\n

Where to get started

\n

\n

Our work is open. Before you apply, spend an evening in our GitHub — it's the fastest way to understand how we think and the strongest signal you can send us:

\n

\n

- github.com/hushh-labs/hushh-research — the product workspace: personal AI experiences, consent flows, and applied agent systems (TypeScript)

\n

- github.com/hushh-labs/consent-protocol — the consent-first Python backend for personal data agents (FastAPI, Google ADK, Supabase)

\n

\n

Read the code. Open a sharp issue. Or better send a PR. The best applications we've seen started there.

\n

\n

Exploring a different seat? We have over 50+ open roles across user experience and design, lead generation , brand and product marketing, core engineering, forward-deployed, systems and AI/ML research, hardware design at our HQ in Kirkland, WA or for one of our Super Intelligence Labs University View email address on us.fitly.work internships and coops available to PhD, Masters and Bachelors students: careers.hushh.ai/open -roles

\n

\n

How we hire

\n

\n

Show us what you shipped, what broke, how you fixed it, and why people keep using it. One honest application a human reads every word: View email address on us.fitly.work

\n

\n

Ownership, consent, and security are non-negotiable. Everything else, we can build together.