Ingeniero de Software — Full-Stack / AI

About Runook

Runook is the Forward Deployed Engineering company for every business that's been priced out of enterprise AI.

Wall Street banks have Palantir. Fortune 500s have OpenAI's Applied AI team. Every other business in America — your local CPA, your favorite Flushing dim sum spot, the freight broker shipping your Amazon order — has been priced out of AI. We're changing that.

Founded by engineers and operators, we run two product lines in parallel:

1. Vertical AI products — AI systems we've shipped in production that customers use every day.
2. An AI deployment service — we embed with traditional SMBs, map their workflows, and ship microservice-based automations that ingest information, analyze it, route notifications, and assign tasks — at 1/10th the cost of legacy consulting firms.

We've already replaced 8 internal headcount with AI workflows we built for ourselves. Now we're scaling that playbook to thousands of customers.

• Stage: Pre-seed / early revenue
• Team: Small, technical, bilingual (English + Mandarin), high-agency
• Customers: NYC / NJ Chinese-American SMBs across logistics, accounting, real estate, beauty, healthcare, F&B

Our values

• Ship before you slide-deck.
• The customer's workflow is the spec.
• "I don't know how" is fine; "I can't figure it out" isn't.
• Bilingual fluency is a feature, not a footnote.

---

About the role

You'll be the engineer who actually ships the systems our Forward Deployed Engineers design with customers. One week you're hardening a freight-pricing engine running in production. The next you're spinning up a microservice that ingests WeChat messages, runs them through an LLM, and routes tasks to a customer's ops team — and watching real revenue land because of it.

We're a small team. You'll own production code from day one. No tickets-from-the-PM-who-talked-to-the-architect game.

What you'll do

• Own end-to-end features across our stack: React/Next.js frontend, Node/TypeScript or Python backend, Postgres, vector DB (Weaviate/pgvector), Redis, AWS.
• Build and maintain the microservice fabric that powers customer deployments — message ingestion (email, WeChat, SMS, web forms), LLM-backed classification, notification routing, task assignment, audit logging.
• Integrate with whatever the customer already runs — DAT, QuickBooks, Yardi, Square, Shopify, Notion, Twilio, Zapier — you'll get good at reading other people's APIs.
• Ship LLM-powered features that actually work in production: RAG pipelines, function calling, evals, output schemas, caching, cost monitoring, fallback chains. Hallucinations are bugs, not features.
• Write code that a future you (or a future colleague) can read at 2am the night before a customer launch.
• Sit in on 1–2 customer calls per week so you understand what you're building and why.

What we look for

Must have

• You've built something real — could be a school project, a hackathon win, an internship deliverable, a side project, or a job. We care that it shipped to someone other than your professor.
• Comfortable across the stack — you can write a React component, design a Postgres schema, and deploy to a cloud without panicking. Depth in one area is fine; total avoidance of any layer is not.
• Hands-on experience with LLM APIs (OpenAI, Anthropic, Gemini) — not just "I tried ChatGPT once." You've at least built a small RAG / agent / multi-step LLM workflow.
• Bias to ship. You write code and finish things.
• Clear written communication — we're async-first, your PR descriptions and Slack writeups matter.

Nice to have

• Bilingual (English + Mandarin) — half our customer base needs Chinese-language UX.
• Experience in one of our customer verticals (logistics, accounting, real estate, healthcare, F&B).
• Open-source contributions or a meaningful personal project portfolio.
• Comfortable with a small ops surface — you've used Docker, GitHub Actions, AWS, and lived to tell about it.

We don't care about

• Whether you have a CS degree. Self-taught shippers are welcome.
• Big-tech pedigree. We care what you've actually built.
• Years of experience. A strong builder with 1 year beats a comfortable 10-year engineer for this role.

You're a fit if

• You'd rather ship a working v0.5 than design a perfect v1.0.
• You read other people's code and learn instead of complaining about it.
• You'd rather debug for 3 hours than ask the same question twice.

Why this role at Runook

• You'll see your code in production within week 1, not month 6.
• You'll learn FDE-style customer thinking from people who've embedded with real businesses.
• You'll work in a modern stack (Next.js, TypeScript, Python, LLMs) without the legacy code archaeology of bigger companies.
• For exceptional contributors: clear path to a full-time founding-team role and senior leadership as we grow.

---

How we hire

Our process is fast (typically 1–2 weeks):

1. Intro chat (30 min) — with a founder. You learn about us, we learn about you.
2. Take-home or live exercise (2–4 hours) — ship a small feature against a real problem we have. We'll send you the spec.
3. Team interview (2–3 hours, in-person or video) — working session with 2–3 people you'd work with day to day.
4. Offer.

We give detailed feedback within 5 business days of any stage. Always.

How to apply

Send the following to info@runook.com (or DM a founder on LinkedIn):

1. Subject line: SWE — Your Name
2. One paragraph: "What's the most interesting thing you've shipped in the last 12 months, and what did you learn from it?"
3. GitHub, portfolio, or resume (whichever shows your work best).
4. Optional but appreciated: a link to a live project we can play with for 60 seconds.

We read every application. We respond within 7 days.