CO-Ver is a self-funded startup building software to help property owners and developers detect and challenge overbilling on construction change orders. Their product uses document ingestion, cost comparison, and LLM-assisted analysis with human experts. They are actively hiring experienced fullstack developers for a contract-to-hire role, indicating a proven need and funding.
What they want, where you stand, and the exact résumé edits to qualify.
Biggest lever: Gain hands-on experience with Convex and Clerk, the specified backend and authentication tools, as they are central to the listed stack.
A starter prompt for Claude Code, what you'll need, and how to reach them.
You are a senior fullstack TypeScript developer. Your goal is to build a web application for CO-Ver (co-ver.com) that helps property owners and developers review and validate construction change orders. The core task is to identify potential overbilling by comparing submitted change order line items against baseline project estimates or typical costs. This will be an MVP on the Lumivara house stack: Next.js 16 App Router, React 19, Tailwind v4, AI SDK v6 with Gemini, Neon Postgres on Vercel. First, focus on the document ingestion and initial comparison workflow. Create a Next.js application. **MVP Slice:** 1. **Document Upload:** Implement a file upload component for PDF and Excel construction change order documents. Use `react-dropzone` for the UI. 2. **Basic Data Extraction (Mock):** For this MVP, simulate data extraction. When a document is 'uploaded', parse a hardcoded JSON mock of a change order (including line items, quantities, unit costs, and total). This mock should represent a simple change order for a small construction task. 3. **Baseline Comparison (Mock):** Create another hardcoded JSON mock representing a 'baseline estimate' for the same construction task. Implement a simple comparison UI that highlights discrepancies (e.g., if a unit cost in the change order is >10% higher than the baseline). 4. **Review Workflow:** Display the comparison results in a simple table. Allow a user to 'approve' or 'flag for review' each line item. Store this state in a Neon Postgres database. Define the schema for change orders, line items, and comparison results. 5. **User Interface:** Use Tailwind v4 for a clean, professional UI. Ensure basic responsiveness. **Technical Requirements:** * Next.js 16 App Router for routing and server components where appropriate. * React 19 for client-side interactivity. * Tailwind v4 for styling. * Neon Postgres for database. Use Drizzle ORM for schema definition and queries. * No LLM integration required for this MVP slice, as data extraction and comparison are mocked. * Authentication (Clerk) is out of scope for this MVP but design the data models to support user ownership of documents. **Build and Verify Gate:** * A user can upload a mock document (handled by the hardcoded JSON). * The system displays a comparison between the mock change order and a mock baseline estimate. * Users can flag individual line items for review, and this state persists in the database. * The application is deployable to Vercel.
United StatesOntario
Focus on specific Ontario construction regulations and local GC practices for change orders, leveraging local pricing data and building codes.
CO-Ver | Fullstack & Senior Fullstack SWE | Remote US | $125k-$175k DOE + Equity | 1099 Contract-to-Hire | Self-funded GCs routinely overbill 10-15% on change orders, and owners have no systematic way to catch it. CO-Ver is building software to help owners and developers review, validate, and challenge construction change orders before they pay them. The product combines document ingestion, scope/cost comparison, estimate review workflows, and LLM-assisted analysis with human estimators in the loop. MVP is nearing completion. We're targeting production by mid-July, self-funded, intentionally small, and working directly with customers in a painful, expensive, and underserved market. You'll work with the founders and fractional CTO, have a real voice in product and architecture decisions, and receive early-stage equity. Open roles: Fullstack SWE ($125k-$155k): 3+ years, strong TypeScript, React/Next.js, solid data modeling, interest in LLM-assisted workflows Senior Fullstack SWE ($150k-$175k): 5+ years, deep TypeScript, production Next.js/multi-tenant SaaS/LLM features, comfortable owning architecture decisions Nice to have: Convex, Clerk, Tailwind v4, PDF pipelines, construction dom
Build a small Next.js + Vercel app (e.g., a simple document processing tool) that integrates Convex for data storage and Clerk for user authentication. Focus on demonstrating how to ingest data (even placeholder text or basic file uploads) and apply LLM processing to it. Target 1-2 weeks for a working MVP.
Standard database for Next.js apps
Standard deployment platform for Next.js
New: authentication provider for multi-tenant SaaS - ~1 day to integrate
Learn it: Auth quickstart (Clerk) ↗
Get set up: Create an auth app (Clerk/Auth0) → copy the publishable + secret keys → add the middleware.
New: fullstack backend as a service, real-time data - ~2 days to learn and integrate deeply
Learn it: Search getting-started ↗
Get set up: Create the account/instance, generate the API key or credentials, and add them to your project's environment variables.
Familiar with AI SDK v6 and prompt engineering for Claude
New: Understanding construction processes, common overbilling practices - ~1 week of research/interviews
Learn it: Search getting-started ↗
New: Working with binary PDF data and extracting text/tables - ~2 days to get robust parsing
Learn it: Search getting-started ↗
Email recruiting@co-ver.com with [hn] in the subject.
“I'm a solo operator specializing in AI-buildable SaaS, with a strong background in Next.js and LLM integrations. I've built an initial prototype for change order review in less than a week — here's a demo. I'm keen to discuss how I can contribute to CO-Ver's mission on a contract basis, leveraging my rapid development skills.”
Open the original ↗