You are the technical partner of the product co-founder, at her level. You own the product end to end from the technical side: the apps, how they meet the hardware, the stack, the releases, and the engineering team. Hardware and logistics sit on a separate line. This is a hands-on role: at the start you write a large share of the app and backend code yourself, and as founding, embedded and LLM engineers join, you move toward validating their decisions while staying close to the code.
The core of the role is mobile apps, in a children's or educational product, ideally from a company people recognise. Speech to text, text to speech and voice commands you need at the level of judgment, not as the job itself. You are the senior person in the room: you look at the system from above, choose a direction, and run the function.
Parent app and backend
- Take the parent app from the hardware line and become its owner. Today that app has no technical owner, and the decisions in it are not validated
- Make the app work for a parent: pair the device, create an account, set controls, see what the child did, delete the data
- Ship iOS and Android with a real release process: review, staged rollout, rollback
- Keep the backend behind those flows working: login, sessions, history, progress, a weak connection and offline
Technical decisions
- Choose the stack and defend the choice
- Validate what the founding, embedded and LLM engineers propose, and own the result
- Decide what runs on the device and what runs in the cloud
Software on the device
- Own the seam with the hardware line: technical requirements, not manufacturing gates
- Hold the quality bar for a four-year-old: the session starts, it holds, it still works when speech is messy
- Judge speech to text, text to speech and built-in voice commands at the level of architecture, not implementation
LLM function
- Manage the LLM engineer and define what good looks like
- Own unit economics as a result: cost per session, and what gets cut when the numbers stop working
Agents
- Own the product agent layer: dialogue, characters, orchestration, evals, and safety agents
- Own how the engineering team works with agents: what the agents do, what stays with people, observability and boundaries
- Build the first version of this layer yourself if needed, then hand it over to the LLM engineer and keep validating it
Privacy and child safety
- Own the technical side of children's privacy: what is stored, retention, encryption, deletion
- Hold the right to block a release on child-safety grounds
Engineering team
- Design the software team, run technical interviews and set the hiring bar
- Sequence the founding, embedded and LLM engineers
- Build the engineering practice: review, ADR, post-mortems, release cadence
- 10+ years as a full-stack engineer with an apps focus: iOS and Android in production is required. You have shipped complete applications end to end: onboarding, account, device pairing, progress, settings, deletion. You still write production code and are ready to carry the app and backend yourself in the first months, before the team is in place
- Experience in children's or educational products: you have built apps where the user is a child and the buyer is a parent or a school. You know how a parent sets up and controls the product, how a child actually uses it, and how a learning product shows progress. Ideally a product people recognise by name
- 3+ years of technical leadership: architectural ownership, engineering standards, review culture, participation in technical interviews. You validate and answer for the result
- Vibe-coding and agents as a regular working mode for the team: UI, API and deployment go out with AI-assisted tools. You control output quality and set what stays with people
- Product judgment: you narrow scope, choose cuts, explain trade-offs to stakeholders, and agree on an acceptable minimum within the timeline
- Confirmed production in mobile and the backend behind it: releases, staged rollout, rollback, incident review, post-release fixes
- Demonstrated speed: cases where you took an app to production or release-ready on an agreed date, without a month of analysis
- Evidence of shipped apps is required: store links, demos, or the named product. We do not need your personal repository
- Children's data in software: what is stored, encryption, retention, a parent can delete it
- English B2: code, documentation, calls, vendors, child-development experts
- An app tied to a physical learning device, not only a standalone mobile product
- Speech to text, text to speech and built-in voice commands at the level of decisions
- Children's privacy in practice: COPPA, Privo or an equivalent process
- LLM tools and agent stacks: LangChain, LangGraph, Mastra, CrewAI, Autogen, function calling, evals - hands-on, and able to set the bar for the team
- A product you took through a hard reset - a shutdown, a pivot, or a rebuild on generative models
- Startups, a venture studio, or a multi-product environment: first hires, first market
- React Native, Expo, Next.js, GraphQL - aligned with the studio portfolio
- Apps: iOS, Android, React Native, Expo, Next.js, onboarding and device pairing, parental controls, progress and account
- Backend: Node.js, GraphQL, REST, WebSockets, OAuth2, JWT, sessions, auth, secrets
- AI and agents: LangChain, LangGraph, Mastra, CrewAI, Autogen, MCP, function calling, structured outputs - hands-on at the start, then architecture and review
- Voice, at the level of decisions: speech to text, text to speech, voice commands
- Device side, at the level of decisions: embedded Linux or RTOS, audio pipeline, Wi-Fi and BLE, signed firmware, OTA
- Observability: OpenTelemetry, Langfuse, Prometheus, Datadog or equivalents
- Practices: Git, CI/CD, Jira, ADR, post-mortems, agent-assisted engineering
- Parent app: pairing, account, controls, history and deletion work in production. No critical defects in the first 7 days after a release, or an agreed hotfix SLA
- On-scope delivery: agreed app and backend definition of done closed by the release date, or a documented fix plan with deadlines
- Continuity: children keep coming back to the learning experience week after week
- Golden samples at the start of December, then the first batch on the agreed date
- Cost per session and response latency p95 inside the agreed budget
- Zero incidents of an unacceptable answer to a child in production. Red-team pass rate above threshold and rising release over release
- Children's privacy for the launch markets closed before shipping, with no findings on children's data
- Share of routine engineering work through agents and vibe-coding (tests, docs, release scripts) as a maturity indicator
- Key engineering roles filled on schedule, with retention in the software team
You stand next to the product co-founder, at the same level. You own the technical function. Hardware and logistics sit on a separate line with their own lead. Manufacturing gates and hardware sign-off are not in this role. You hold the seam and give that line technical requirements.
Reporting into you: Founding Engineer (Software), Embedded Software Engineer, LLM Engineer, and further software hires as the product grows.
Technical decisions are autonomous: stack, architecture, the on-device and cloud boundary, model choice, releases, the agent layer, and the safety veto on a release. With the product co-founder you agree scope, definition of done, and ship date. First-batch volume and cost targets are set jointly with the founders. Price, launch markets and launch order stay with the founders, with input from this role.