AI Full-Stack Developer for inventory management software (LLM features + web + APIs)

About us Acctivate is inventory and order management software for small and mid-sized manufacturers, distributors, and online retailers that have outgrown QuickBooks inventory. Customers use us for real-time inventory, purchasing, warehouse, fulfillment, lot/serial traceability, landed cost, EDI, and multi-channel sales, while they keep QuickBooks for accounting. I am David Huang, VP of Development and CTO. I have been with this product from the beginning. We are a focused product team. We want practical AI that helps operators make better inventory and order decisions, not demos that look impressive and fail on real data. The work We need an AI full-stack developer to design, build, and ship production features that combine a modern web/API stack with LLMs and related AI services. You will work against a mature operational product and real business data: items, warehouses, purchase orders, sales orders, lots/serials, costs, and QuickBooks sync. This is not a chatbot-only gig and not a greenfield toy app. The work has to be accurate, auditable, and safe to put in front of customers who run warehouses and accounting from this system. Initial scope (we will refine after a short paid trial) - Design and build AI-assisted product features (examples: intelligent search, document extraction from POs/invoices/packing lists, anomaly flags on inventory or orders, assisted Q&A over product/help content, workflow suggestions) - Build the full-stack path: UI, API, orchestration, evaluation, logging, and fallbacks when the model is wrong - Connect AI services to existing product data and APIs without corrupting inventory or accounting data - Add evaluation, prompt/version control, and basic observability so we can see quality over time - Document how each feature is verified and what it must never do automatically What you will do - Own features end-to-end: UX, backend, data access, model/tool calls, and deployment - Use LLMs with tools/RAG where needed, not as a black box over private operational data - Keep human review on any action that changes inventory, orders, costs, or accounting - Write clean TypeScript/JavaScript and/or Python - Work with SQL data and existing APIs - Ship small, reviewable increments - Communicate tradeoffs and blockers in writing Must-have experience - 4+ years professional full-stack development - Shipped at least one production feature that used LLMs or similar AI APIs (not only tutorials) - Strong TypeScript or JavaScript (React or similar on the front end; Node or equivalent on the back end) - Ability to design APIs, auth, and data flows that are safe for business software - Practical prompt design, tool/function calling, and RAG - Clear written English - Overlap of at least 3 hours with U.S. Central Time Strong plus - Python for AI/data services - SQL Server or other relational databases used in production - Inventory, ERP, WMS, eCommerce, or accounting domain experience - QuickBooks, EDI, or marketplace integrations - Evaluation harnesses, tracing (for example LangSmith or similar), and cost controls - Azure or AWS deployment experience How we work - Git-based workflow, small PRs, code review - Hourly to start, with a 1-2 week paid trial on a real slice of work - If the trial goes well, ongoing freelance work - You will work directly with the CTO - We prefer boring, reliable architecture over extra frameworks To apply, do not send a generic cover letter. In your proposal include: 1. 4-6 sentences on a production AI feature you built: the problem, the stack, and how you handled wrong answers. 2. Your primary front-end, back-end, and AI tools. Say what you do not want to work on. 3. Typical weekly availability (hours) and timezone. 4. Hourly rate and earliest start date. 5. One example where an LLM would have been dangerous in a business workflow, and what guardrails you used. 6. Links to GitHub, a live demo, or a short description of a relevant private project. Applications that only say "I am an expert in AI and full-stack, let's talk" will not be reviewed. Budget and terms - Contract: independent contractor - Engagement: hourly - Start: as soon as a strong fit is identified - Duration: 1-2 week paid trial, then ongoing if successful To show you read this post, start your proposal with the word "Sports" as the first word on the first line. Then continue with the 6 items listed above. Proposals that do not start with Sports will not be considered.