$215,000–$235,000 base salary, plus 0.25%–0.5% equity. Up to $10,000 relocation support.
We are looking for a Founding AI Product Engineer with 3–6 years of experience to own and productize the core backend platform at Arzana — a YC-backed AI company automating the front office of American manufacturing. You'll be one of four or five engineers who build the agent runtime, extraction frameworks, and integration infrastructure that powers all 22+ customers, not just one. This is a greenfield role with real ownership: you define what becomes part of the platform.
What will you be doing?
Build and own a core domain of the backend platform — agent runtime, extraction pipelines, ERP integration frameworks, or evals — with your work shipping to all customers
Productize and unify existing solutions: take six overlapping order entry tools with 20% shared code and get that to 80–85% reuse
Pair with Forward Deployed Engineers on customer requirements that have no framework yet, and decide what gets built into the platform
Ship fast — new customers go live in under a week, and you're one of the people who makes that true
Occasional travel (~5 days/month) to customer plants to watch real users interact with what you've built
Key Requirements
Production AI systems experience — multi-step agent pipelines, RAG, VLM extraction, evals running for real users
Strong Python and real architectural range across backend, data, and cloud (AWS); full-stack capability a plus
Proven end-to-end product ownership — you can point to shipped products and explain every decision
CS or STEM degree (Bachelor's minimum); founder, co-founder, CTO, or founding engineer experience strongly preferred
U.S. person as defined under 22 CFR 120.62 (ITAR compliance required); based in or willing to relocate to New York
Arzana is an AI platform company rebuilding the front office of American manufacturing — automating order entry, quoting, RFQ processing, purchasing, AP/AR, and customer service for manufacturers who still run on fax machines and decade-old spreadsheets. Founded by Marshall Kools and William Alexander, two Stanford engineers and third-generation manufacturing families from Wisconsin and Iowa, the company was born from a deeply personal frustration: watching family businesses spend hours every day on manual data entry and slow quoting cycles. Arzana graduated from Y Combinator's P26 batch, raised a $4.3M seed round, and has grown 30% month over month — 10x-ing revenue during YC alone. Today, Arzana serves 22 customers ranging from 20-person shops to Fortune 500 manufacturers, and has processed over 100,000 purchase orders. The platform deploys semi-custom AI agents built primarily in Python, running on isolated AWS instances so no customer data leaves for outside providers, and already integrates with MES and computer vision systems. The team is seven people strong, headquartered in New York, and operates with a high-ownership, fast-moving culture where every engineer leads their own projects end-to-end — from client communication to deployment. The founders' long-term mission is to fully automate 10,000 factories by 2035, starting with order entry and quoting and expanding to full office and floor operations.
Team size
8 employeesFounded
Website
arzana.comTotal funding
Company locations
Arzana is a small, tight-knit team of eight to nine people — all young, all hungry. The oldest person on the team is 25. The two founders, Marshall Kools and William Alexander, are Stanford engineers and third-generation manufacturing families from Wisconsin and Iowa. They started the company because they lived the problem — watching family businesses drown in manual data entry and slow quoting — and they built Arzana to fix it. Between them, two of their first three companies exited before this one.
The culture is defined by autonomy, ownership, and speed. Every engineer owns their own project end-to-end — you're talking to clients, writing the code, deploying it, and iterating based on real feedback. There's no product manager writing your tickets and no roadmap telling you what to do next week. You move as fast as you want to, with no administrative bottlenecks in the way.
The team values self-starters who build things because they care about the problem, not because they want the title. Marshall puts it plainly: did you start something because you wanted to solve someone's problem, or because you wanted to be a founder? That distinction matters here. The engineers they've hired and promoted fastest — including one who went from new hire to senior engineer in under three months — are people who show up with positive energy, think out loud, explain their reasoning, and take real ownership.
The office is in New York City, and the team is in five days a week. The founders are always there. Weekends aren't required, but about half the team comes in anyway because they're genuinely excited about what they're building. When engineers aren't in the office, they're often on-site with clients — visiting plants, running implementation sessions, and sitting with the people who actually use the tools they ship.
Promotion is fast and merit-based. Arzana has a structured performance framework tied to client communication, deadline delivery, and project quality — so engineers always know where they stand and how to move up. The team is growing quickly (at least two hires a month), which means real opportunities to grow into leadership as the company scales toward its goal of automating 10,000 factories by 2035.
Arzana builds AI agents that automate the front office of manufacturing companies — from order entry and quoting to RFQ processing, purchasing, AP/AR, and customer service. American manufacturing still runs on paper: distributors take orders by fax, key them into ERPs by hand, and quote off spreadsheets built two decades ago. Arzana replaces that manual workflow with purpose-built AI.
The platform is primarily Python-based software (~85–90%) with semi-custom, fine-tuned models trained on domain-specific manufacturing data. Each deployment runs in an isolated AWS instance, meaning no customer data ever leaves for an outside provider. Arzana integrates with MES systems, computer vision platforms, and major ERPs like Epicor — turning complex, bespoke manufacturing workflows into configuration problems rather than custom engineering projects.
Key capabilities include:
Automated order entry — ingesting orders from fax, email, and other legacy inputs and posting line items directly into a customer's ERP
AI-powered quoting — training models with deep geometric and domain context to generate accurate quotes for complex manufactured parts (e.g., mold and tool manufacturing)
RFQ processing, purchasing, and AP/AR automation — eliminating manual data entry across the full manufacturing front office
Agentic extraction and evaluation frameworks — multi-step agent pipelines, RAG, and VLM extraction that run reliably in production for real users
Arzana serves 22 customers ranging from 20-person shops to Fortune 500 manufacturers, has processed over 100,000 purchase orders, and is growing 30% month over month. The long-term vision is to fully automate 10,000 factories by 2035 — starting with the front office and expanding to full plant-floor operations.