"We built 3 AI agents in 48 hours. You'll help us build the next 10."
Matt Putra, Founder. Three agents are live in production today, wired to Slack, Fireflies and Meta.
The build task is paid. We pay you for your time even if we don't move forward.
Why this is a rare seat: most engineers are still waiting for permission to use AI at work. Here, it's the whole job. You'll ship real agents into production, learn directly from our founder Matt Putra (a former PE investor who builds in Claude Code daily), and grow as fast as our AI practice does. We even pay you for the 30-minute build task we use to test fit.
Eightx is a growth-economics consultancy that serves ecommerce, DTC and CPG brands doing $2M–$150M in revenue. We act as fractional CFOs and growth advisors who sit inside our clients' businesses, running financial models, managing ad performance, and tracking every deliverable.
We're now building an internal AI agent system that automates the work our team does every day: reading call transcripts, updating financial models, monitoring ad performance, tracking client deliverables. Three agents are already live in production on a self-hosted VPS, wired to Slack, Fireflies and Meta. You'll be the engineer who builds everything next.
You'll work directly with our founder, Matt Putra, to build, ship and maintain the AI agent infrastructure. This is not a research or architecture role. It's a build role. We move extremely fast. You'll take a spec, pick it up in Claude Code, and have something working the same day.
This isn't a vague "ramp-up." A realistic first deliverable: ship an MCP server that wires one of our data sources (say, Fireflies call transcripts or a Google Sheet financial model) into an agent that posts to Slack, where "shipped" means it runs on the VPS and the team uses it. From there you'll build out four secure agents (we'll show you how, and you'll deliver), replicate a couple client-side, and start enabling the team. The specs exist. The VPS is running. You're shipping into production from day one.
We don't want a cover letter. We want to see you work. When you apply, send us:
If your profile looks right, we'll send you a paid test task: build a working MCP server against a real API from our stack. You'll have 30 minutes. That's not a typo. We pay you for your time regardless of outcome.
As part of early screening, we'll also ask for a short 2–3 minute video (phone or webcam is fine) on why you're interested and your key strengths for the role.