**Type**:Full-time | Founding team | IC with tech lead scope
**Stack**: TypeScript, Next.js, Node, Vercel/Edge, queues, Mongo/Redis, OpenAI Agents, RAG
**About Us**
We're a VC-backed startup changing how humans work with AI.
We're building the first truly full-stack, extensible, and easy-to-use AI backend platform for SaaS teams to integrate deep agents (think Cursor-class) into real products—securely, observably, and at scale.
If Vercel and Supabase had an AI infra child built for agentic apps, that's us.
We ship a cloud platform and OSS packages/demo apps so developers can get from zero → production fast.
**The Role**
We're hiring a senior-to-lead full-stack engineer who is an exceptional TypeScript builder with deep AI engineering intuition.
You'll be the founding IC who both ships core features and sets engineering patterns, mentoring others by example (not management heavy).
You'll lead the build-out of:
- Our open-source demo app & packages (editor plugins, SDKs, adapters)
- Our multi-tenant cloud (serverless APIs, queues, webhooks, RAG/Agents, review workflows)
- Tooling and guardrails that make deep agents safe, fast, and composable
Bring opinions.
Help define the primitives thousands of SaaS companies will rely on.
**What You'll Do**
**Own from spec → prod**: Design and ship high-performance TypeScript services on Vercel (Edge/Node) with durable patterns (idempotency, retries, observability).
**Build the agent stack**: Wire OpenAI Agents/Assistants + tool calling, review queues, diff/suggestion systems, retrieval (RAG), vector pipelines, and cost/latency guardrails.
**Queues & events**: Model jobs with Upstash/Redis queues, cron, webhooks, backoff/retries, exactly-once semantics where it matters.
**Data & multi-tenancy**: Implement clean row-level isolation, optimistic concurrency,



and audit trails across MongoDB Atlas (serverless).
**Performance & reliability**: Profiling, tracing, circuit breakers, caching strategies, and graceful degradation paths.
**Security by default**: Secrets management, scoped auth (JWT), HMAC webhooks, abuse controls.
**Set the bar**: Code reviews, architecture docs, crisp PRs, and hands-on mentoring of incoming engineers.
**Example problems you'll tackle (month 1-3)**
Stream agent tool calls into a real-time **Suggestions & Review Queue** with idempotent apply/undo and atomic versioning.
- Ship a nontrivial **RAG pipeline** (chunk → embed → upsert → retrieve) with per-tenant filters and cost caps.
Abstract out for graphRAG, episodic memory etc.
- Design a **schema-driven adapter layer** so agents can safely edit rich text (Tiptap/ProseMirror) and tables.
- Make a **one-click OSS demo**: Next.js app with embedded editor/table, chat panel, and live suggestions.
**Minimum Qualifications**
**7+ years** building modern web backends/frontends with **TypeScript** (Node + Next.js).
You've owned prod systems at scale.
**Proven full-stack chops**: API design, data modeling, React/Next UI, and performance tuning.
**AI engineering experience**: agents/tools, LLM orchestration, retrieval workflows, prompt/tooling design, cost/latency tradeoffs.
**Distributed/system thinking**: queues, eventual consistency, idempotency, retries, backpressure.
**Cloud/serverless**:



Deployed and scaled apps on Vercel/AWS/GCP, comfortable with edge vs. region tradeoffs.
**Strong product sense**: you can turn fuzzy specs into sensible primitives and crisp UX.
**Badass track record**: OSS, papers, talks, or gnarly systems you can walk us through.
**Nice to Haves**
- RAG/vector experience (e.g., Turbopuffer, pgvector, Milvus, Pinecone).
- Editor tech (Tiptap/ProseMirror) or diff/OT/CRDT familiarity (Y.js).
- MongoDB Atlas (serverless) and Redis/Upstash at scale.
- AI/Python practical experience; maybe you've read Chip Huyen's AI Engineering or Norvig's AI: A Modern Approach, or maybe you have a ML background.
- Observability (OTel, traces/metrics/logs), security reviews, cost controls.
**Our Tech (Current & Near-Term)**
**Frontend**: Next.js (App Router), React, shadcn/ui, Tiptap, TanStack Table.
**Backend**: Vercel Functions/Edge (TS), Webhooks, Upstash Q/Redis, Vercel Cron.
**Data**: MongoDB Atlas (serverless, multi-tenant row isolation).
**AI**: OpenAI Agents/Assistants API, embeddings, tool calling, structured suggestions.
**RAG**: Turbopuffer vectors, scope-aware filters, chunkers.
**DX**: Type-safe SDKs, OpenAPI/TS types, example apps, templates.
**How We Work**
**Default to action.
** Small PRs, daily shipping, measurable outcomes.
**IC ownership.
** You'll lead by building—set standards, not meetings.
**High bar, kind team.
** We give crisp feedback, write clear docs, and leave things better than we found them.
**Open by default.
** We OSS what helps the ecosystem.
**Compensation & Upside**
Competitive salary + meaningful equity (founding-tier).
New hardware, learning budget, and conference support.
Projected Total Compensation: $1,500,****** - $1,850,0

📌 Senior Ai Scientist / Staff Engineer - Founding Ic (Xico)
🏢 ARCOS Labs
📍 Xico