Artificial Intelligence Engineer

About NodePlus

NodePlus is an enterprise artificial intelligence operating system built for regulated industries. We are an application factory with a connector layer architecture that lets large organizations ship AI-native software at the speed of a startup and the rigor of an enterprise. Our live proof of concept is Kaycha Labs, a nine-state cannabis testing network, where NodePlus runs production workloads across operations, sales, compliance, and finance. We are a small team building a serious platform. We are funded, post-revenue, and currently raising a $50 million round to scale.

The Role

We are hiring our next AI Engineer. You will own features end-to-end, from data model to deployed UI, with a particular emphasis on the AI systems that make NodePlus a category-defining product. You will work directly with the CEO/CTO and have outsized influence on our architecture, our hiring bar, and our product direction. You will ship in week one. You will own systems that customers depend on. You will build the evals before the features.

What You Bring:

Core Engineering

• Python 3.11 or newer, with FastAPI, async patterns, Pydantic, and pytest
• TypeScript with Next.js 16 App Router, React 19 (including use() and Actions), Tailwind 4, and shadcn/ui
• SQL and Postgres at depth, including schema design, Supabase Row-Level Security, migrations, window functions, and pgvector
• Cloudflare Workers, including edge functions, Durable Objects, KV, and R2
• Git and GitHub workflow discipline, trunk-based development, clean PRs, and CI via GitHub Actions

AI and LLM Systems

• Production experience with LLM APIs across Anthropic (Claude 4.x), OpenAI (o-series), Google (Gemini), xAI (Grok), and DeepSeek
• Prompt caching, tool use, structured output, and streaming via Server-Sent Events at production grade, not demo grade
• RAG systems built end-to-end: chunking strategy, embeddings, metadata filtering, reranking, and evaluation
• Vector databases, with ChromaDB as primary and pgvector as secondary
• Embeddings model selection (such as snowflake-arctic-embed2) with a clear understanding of cost-quality tradeoffs
• Ollama deployment at scale, including model quantization, context window tuning, and multi-GPU layouts
• Fine-tuning with QLoRA or LoRA via Unsloth or Axolotl, including dataset curation
• Model Context Protocol (MCP), including writing servers and registering them with Claude Code or Claude Desktop
• Evaluation discipline: offline and online evals, with the standing rule that no AI feature ships without one

Infrastructure and Operations

• Docker and docker-compose, Linux services via systemd, Windows services via NSSM and PM2
• PowerShell 5.1 and 7, plus bash. Our fleet is Windows-heavy with Linux servers; you need to be fluent on both
• Tailscale or WireGuard, SSH mesh networking, MeshCentral helpful
• Monitoring with Prometheus, Grafana, or equivalent

Product Sense

• You can ship a working feature end-to-end without hand-holding
• You write the eval before the feature
• You know when not to use an LLM, because deterministic code beats a prompt
• You debug hallucinations by fixing retrieval, not by stacking more prompts

How We Work

Small team. Direct communication. Trunk-based. We measure shipping velocity and customer outcomes, not lines of code. We use AI tools aggressively in our own workflow because we believe in our product. We have strong opinions, weakly held, and we change them when the data says so.

What We Offer

• Meaningful equity in a company with real revenue and a serious enterprise wedge
• Direct access to the CEO/CTO and to customers
• Modern tooling and the budget to use it
• The chance to build the operating system that the next generation of regulated enterprises will run on

How to Apply

To apply, email our CEO directly at james@nodeplus.ai and cc careers@kaychalabs.com with your GitHub, a link to something you have shipped that you are proud of, and one paragraph on the AI feature you would build first if you joined. We read every application and respond within five business days.