The role
You'll be one of the first engineers dedicated to the Intent Refinery product line. You'll work directly with the Head of Product and the CEO. There is no backlog to groom — there's a revenue loop to close, and you'll have the autonomy to close it.
End-to-end means exactly that. We're a team of under 30. There is no DevOps team, no SRE team, no platform team. You own the full lifecycle of what you build — from API design to frontend experience to deployment to observability to the alert that tells you something broke at 2am. This isn't a staffing gap; it's how we believe software should be built at this stage. If you've only ever worked in organizations where someone else handles deployment, this will feel uncomfortable. If you've been wanting the kind of ownership where your code ships to millions of users and you can trace the revenue impact the same week — this is it.
What you'll actually do
Close the loop between intent capture and revenue. This is the job. Everything below serves this mission.
- Design and build backend services (Go / Python) with an OpenAPI-first approach — our APIs are evolving toward MCP-compatible agent interfaces where machines, not just humans, consume our product
- Build frontend experiences (Vue.js / Nuxt.js) that capture user intent: AI-powered Q&A widgets, conversational flows, and rich media interactions embedded across publisher sites
- Integrate LLM APIs and AI models into production features — intent classification, sponsored question matching, conversational commerce — at real scale, not as experiments
- Rapid-prototype new AI experiences and kill or ship them within days, not months. We produced a full AI animation pipeline in 45 minutes using 46 API calls chained together. That's the velocity we expect.
- Deploy, monitor, and operate what you build — Kubernetes (EKS), Prometheus, Grafana, Sentry. You're the first responder because you're the person who knows the system best
- Treat cloud cost as a product constraint — we operate across GCP, AWS, and multi-cloud CDN architectures, and FinOps discipline is part of engineering excellence here
- Use AI tools as your default workflow. Claude Code, Cursor (company-provided), and internal code generation aren't optional — they're how we expect you to work. If you're still writing boilerplate by hand, you're moving too slowly.
What we're looking for
The non-negotiables:
- You've shipped production SaaS products end-to-end — frontend, backend, deployment, monitoring. Multiple times. In environments where that's not my job wasn't an option.
- You're comfortable in Go or Python on the backend and Vue.js on the frontend. You don't need to be an expert in all of them today, but you need to be productive in weeks, not months.
- You've integrated LLM APIs or AI models into production features that real users touched.
- You think in API contracts and system interfaces, not just code. You design for the next team member who hasn't been hired yet.
- You default to shipping. When the brief is messy, the requirements are half-formed, and nobody's drawn a wireframe — you make a decision, build a version, put it in front of users, and iterate. Waiting for clarity is not in your vocabulary.
What would make you exceptional:
- Experience in ad tech, martech, or publisher monetization — you understand why intent data is worth more than impression data, and you know what CPM, CPC, and ROAS mean in practice
- Familiarity with agent frameworks and protocols (MCP, A2A, LangGraph, n8n) and a point of view on where AI-native software architecture is headed
- You've worked at a company where the product you built directly determined whether the company raised its next round or signed its next customer
- 中文溝通能力 — 我們團隊日常混用中英文,如果你能順暢地用中文討論技術方案、參與產品討論,會是很大的加分
Tech stack
- Frontend - Vue.js / Nuxt.js / Vuex
- Backend - Golang / Python / PostgreSQL / Redis
- Infrastructure - EKS (Kubernetes) / GCP / AWS / multi-cloud CDN architecture
- Observability - Prometheus / Grafana / Sentry / Loki
- AI tooling - Claude Code / Cursor / internal code generation tools
- Data - Databricks / S3 clickstream pipelines / vector search
Our architecture spans content delivery optimization, real-time behavioral data capture, and AI model orchestration across a multi-cloud footprint. If genuine systems complexity is what excites you — not toy problems — you'll find plenty here.
Want to see how we actually build this
→ How we ship: Becoming Product Builders with Business Thinking
