Fractional AI Engineer — an FDE who ships, on your team, at half the rate.
A US-based specialist fractional CTO charges $300-500/hr. Most SMBs can't sustain that. We sit in the middle slot: an embedded engineer who is deep enough to pick the right model, write the tool contracts, and merge the PR — at $2-5k/month on EU hours.
The problem
You have a product team, an ops team, and a "someone should do AI" problem. You can hire a full-time AI engineer for $180k/yr in Berlin or $300k+ in SF. You can buy consulting decks for $30k. Both options move the needle slowly.
A fractional AI engineer ships features every week, pairs with your real engineers, and leaves behind documentation so you eventually don't need us. It compounds without the burn rate.
What's included
- One or two production PRs per week on your repos.
- Weekly 30-min pair session (async-friendly).
- Slack/Telegram access with one-business-day SLA.
- Model + prompt + infra decisions documented in-repo.
- Direct introduction to every dependency we bring in.
- Monthly retro — what compounded, what didn't, what to kill.
Good fit / bad fit
- • Live product, paying customers, AI is a missing feature.
- • Team < 15, one or two engineers, shipping weekly.
- • Needs one clear AI capability in production, not R&D.
- • Comfortable with async + 2-3h EU-timezone overlap.
- • Pre-product, pre-customer, pre-PMF.
- • Expecting 40h/week attention for $2k/mo.
- • Only wants strategy or deck output.
- • Hardcore US-only timezone with same-day response SLAs.
FAQ
How is this different from an AI consultant?
Consultants produce recommendations and decks. A Forward-Deployed Engineer ships production code and pairs with your team. Our retainer exists because SMBs don't need strategy — they need someone who merges PRs on Friday.
What does a typical week look like?
Monday 30-min async standup. Wednesday pair session. One PR shipped by Friday. A Telegram/Slack channel that gets a reply within a working day. Async-friendly, EU-timezone overlap for live calls.
What's the minimum commitment?
One month minimum, then month-to-month. We don't do 12-month lock-ins — if we're not compounding value by month two, either side should walk.
Can you onboard our in-house engineer at the same time?
That's the best version of this engagement. We leave behind an engineer who can keep shipping AI features without us. The retainer is designed to ramp down, not up.
Need a shipping AI engineer next week?
Book a 20-min call. Bring one feature you'd want merged to main. We'll tell you if it's a 2-hour or 2-sprint job.