Ship your backlog 2–3x faster. I security-review every PR before you see it.
You pick the tickets and I fix the price before we start. AI agents write the code. I review each PR line by line, security first, then send it to you.
If the first reviewed PR isn't merge-ready to your standard, you don't pay for it.
10+ years shipping production software with
- Bono Fintech
- GoSocial
- PGL Esports
- Ipsos
30 reviewed PRs in under 3 days · 3 shell-injection paths caught before merge · every miss listed in the case study
Prices
Start with one Reviewed PR at €1,000, or go straight to a sprint. Move up a tier within 30 days and what you already paid counts toward it. All prices are in EUR. Prices exclude VAT; VAT is added where it applies.
PlanA written build plan, or ranked findings with the top fix shipped
Architecture Session
1 h
One hour on what you want built. Within one working day you get a written build plan: architecture, tooling, cost estimate. If we start the build within 30 days, the €200 counts toward it. Fixed price, paid up front.
€200
AI Code Audit
2–3 days
Your AI-written or vibe-coded code, LLM workflow or token spend, reviewed: security gaps, cost per merged PR, cache use and wasted agent runs. You get the findings ranked by what to fix first, plus one PR that fixes the top finding, scoped on the findings call. Price agreed before I start.
from €1,500
Set upYour team's rules file and agent setup, running on your repo
Team workshop
3.5 h / 8 h
Claude Code, parallel agents, a rules file and review gates, hands-on. Your team leaves with its first rules file and a worktree-per-agent setup running on your repo. Up to 10 people. Fixed price for up to 10 people, agenda agreed in advance.
€3,000 half-day / €5,000 full-day
BuildOne automation or agent setup, built and handed over
AI Automation
1–3 wk
One n8n workflow or LLM pipeline, built with evals and cost controls from the start. Where it writes for customers, it gets a gate that stops it inventing numbers or facts. Fixed price before I start; evals prove it works before handoff.
from €3,000
Agent Fleet Setup
1 wk
The harness I run (the rules, scripts and CI checks that keep agents on track), with parallel agents in git worktrees, installed in your repo and handed over to your team. Price fixed for one repo before I start.
from €4,000
ShipYour backlog, shipped and reviewed
Reviewed PR
~48 h
One real ticket from your backlog, shipped by an agent and security-reviewed line by line.
€1,000
Delivery Sprint
1–2 wk · larger scopes 2–3 wk · One at a time
A fixed batch of merge-ready, security-reviewed PRs at a price agreed before we start: €950–1,500 per PR, security report included.
from €7,500
Retainer
monthly · One at a time
A steady monthly flow of reviewed PRs, with a review slot reserved for you.
from €4,000/month
I review every PR myself, so I run one sprint or retainer at a time. If I'm booked, I'll tell you the next start date.
How a sprint works
Four steps. Scope and price are fixed on day 0, so there's no runaway bill.
- 1 · Scope
Scope call (day 0)
We agree the PRs, what done means for each, repo access and your conventions. Scope and price are fixed here. If it isn't a fit, you'll know on the call.
- 2 · Ship
Agents write the code (days 1–N)
Each ticket goes to a Claude Code agent in its own git worktree, following your CLAUDE.md. Several run in parallel, which is where the 2–3x comes from.
- 3 · Review
I review every PR (continuous)
Security first (injection, auth, secrets, access control), then correctness and scope. I assume the agent got it wrong until the code shows otherwise. More agents don't speed this step up, which is why I take one client at a time. How the review works →
- 4 · Handoff
Handoff (final day)
You get the reviewed PRs on real branches in your repo, the security findings for the whole sprint, and a note on what the agents kept getting wrong. You merge.
Main offer
Delivery Sprint
A fixed batch of production-ready pull requests. AI agents write them, and I review each one line by line, security first. You agree the price before we start.
What you get
- A fixed number of merge-ready PRs on real branches in your repo, reviewed and ready for your team to merge.
- A review note per PR: what the agent wrote, what the review caught and changed, and any risk left.
- The security report, yours to keep: every vulnerability class checked, and everything found and fixed.
- A list of what the agents got wrong in your codebase, so your team knows where to look from now on.
Sprint prices
A Standard Sprint is €14,500 for 10–15 PRs, each one security-reviewed, plus the security report.
- First sprint
Pilot Sprint
- Best for
- A first engagement, to see how we work together
- Roughly
- A focused feature slice (~5–6 PRs)
- Timeline
- ~1 wk
- Price
- €7,500
- Best per-PR price
Standard Sprint
- Best for
- A steady backlog, cleared in one or two weeks
- Roughly
- A reviewed contractor-month (~10–15 PRs)
- Timeline
- 1–2 wk
- Price
- €14,500
Scale Sprint
- Best for
- A larger surface, with more PRs and deeper review
- Roughly
- A major surface, in depth (~16–20+ PRs)
- Timeline
- 2–3 wk
- Price
- from €24,000
| Feature | First sprintPilot Sprint | Best per-PR priceStandard Sprint | Scale Sprint |
|---|---|---|---|
| Best for | A first engagement, to see how we work together | A steady backlog, cleared in one or two weeks | A larger surface, with more PRs and deeper review |
| Roughly | A focused feature slice (~5–6 PRs) | A reviewed contractor-month (~10–15 PRs) | A major surface, in depth (~16–20+ PRs) |
| Timeline | ~1 wk | 1–2 wk | 2–3 wk |
| Price | €7,500 | €14,500 | from €24,000 |
| Book a scoping call | Book a scoping call | Book a scoping call |
The exact price is fixed on the scope call, before any work starts.
Guarantees
- Security
- Every PR gets a documented security review before you see it. If an in-scope vulnerability later turns up in a PR I marked reviewed, I fix it free and re-review the batch around it.
- First PR
- If the first PR isn't one you'd merge, you don't pay for it. Your risk is highest before I've shipped anything.
What this is not
- Open-ended work. The PR count and scope are fixed on day 0, and new scope is a new sprint.
- A staff-augmentation seat. You buy reviewed PRs, and nobody joins your standups.
- Greenfield architecture. Designing a new system from scratch is a separate conversation.
- Auto-merge to production. The final merge and deploy stay with your team, so your change control stays intact.
- Ongoing maintenance. On-call and upkeep after handoff belong to the retainer.
Payment
50% to book the slot, 50% on handoff. A Pilot Sprint can be paid in full up front. Running sprints back to back? See the retainer below.
“A careful analysis. He caught the contradictions in the handoff before any code was written, and his corrections were genuinely useful.”
Not ready for a sprint?
One Reviewed PR: €1,000
One ticket from your backlog, shipped and reviewed, before you commit to a sprint.
- You send one real ticket: a bug or a small feature that's already queued.
- An agent writes it. I review the diff line by line, security first, the same pass every sprint PR gets.
- You get the merge-ready PR in your repo, plus a one-page report of what the review caught.
- Delivered in ~48 h.
Retainer
Reviewed PRs every month, at a pace we agree up front. Your review slot is held for you.
What you get each month
- An agreed number of PRs, each one reviewed the same way as in a sprint.
- A running log of security findings, and a monthly note on what the agents got wrong in your codebase.
- Async access to me for delivery and review questions.
- First claim on my review time: retainer clients keep their slot ahead of one-off sprints.
From
€4,000/month
3-month minimum, from €12,000
The monthly PR count and per-PR rate are set on the call, against your repo and pace. The per-PR rate lands about 20–30% below a one-off sprint.
Guarantees
First-month guarantee. We agree the pace up front. If month 1 misses it, you can stop there and owe nothing more.
Every retainer PR also carries the sprint's security guarantee.
Terms
- Monthly volume has a cap. Overflow moves to next month or a top-up sprint.
- 24/7 on-call isn't included. It can be added as a separate line item.
- Three months to start, then month to month with 30 days' notice.
“Whatever I flagged, he handled right away. Tasks disappeared as fast as I could add them.”
Agent Fleet Setup: my agent setup, in your repo
I install the harness I run in your repo, tuned to your stack and conventions, so your own team can run agents with the review gates already in place.
What you get
- A working Claude Code setup on your repo: project rules in CLAUDE.md, one worktree per agent, and CI wiring.
- The gates I run on my own work: a PR can't be opened until typecheck, lint, build and tests pass, and a commit with a secret in it is refused.
- My security review checklist, the one that caught 3 shell-injection paths on my own 30-PR run, adapted to your stack.
- A hands-on session: I run a live batch on your backlog with your team, then your team runs the next one while I review.
- A written playbook: how to hand out tickets, what to review first, and where a human check is required.
- 2 weeks of async support after handover.
From
€4,000
One repo: agent setup, CI security gates, review checklist, playbook and support. More repos, extra workshops or pairing raise the scope, and we fix the price on the call before any work starts.
Guarantee: Working setup. By the end, your team has shipped at least one reviewed PR through the setup themselves, with me present. If they haven't, the sessions continue free until they have.
Terms
- Setup and handover only. For ongoing delivery, see the sprint or the retainer.
- Your team takes part in the sessions, because the goal is that they can run it without me.
- Claude Code licences and other tool seats are yours to bring.
- I don't keep access after handover unless you add a retainer.
Questions
How good is the code if AI agents wrote it, and what if the agents get it wrong?
Agent output is a draft. Every PR gets a line-by-line review that assumes the agent got it wrong until the code shows otherwise. When an agent can't do a ticket, I fix it or write it myself: a failed run costs me time, and the price covers the reviewed result. Agents give you 2–3x throughput and get specific things wrong, so every sprint ships a log of what the review caught in your codebase.
Unsupervised agents ship vulnerabilities. How do you handle security?
They do. Veracode's July 2026 test found a known vulnerability in about 44% of AI code-generation tasks. That's why security is the first pass of every review: injection, auth, secrets, access control. On my own 30-PR run, a repo with no test suite, the review caught 3 shell-injection paths from request input before merge. Anything in scope that I miss is covered by the security guarantee in the sprint terms.
Who owns the code, what happens to our data, and will you sign an NDA?
You own everything from the first commit. I sign your NDA or DPA before I get repo access and work in your repository under your access controls. Which AI tools touch your code is agreed in writing first; if your policy rules one out, it doesn't run on your repo. Your code is never reused for another client or published, and my access ends at handoff unless you're on a retainer.
Which stacks and languages do you review?
TypeScript and JavaScript end to end: Node, Nest.js, React, Next.js, Tailwind, SQL and MongoDB. That's where 10+ years of production work went. The security pass (injection, auth, secrets, access control) applies to any language, but if your stack is outside that list I say so on the call.
What timezone do you work in?
Romania (EET/EEST, UTC+2/+3), which overlaps fully with European hours and with the morning of US East Coast teams. Day to day runs async in writing: you get the PRs, the review notes and a short written update, so nobody waits on a meeting.
What counts as one PR?
One ticket from your backlog, shipped as one reviewable pull request: a bug fix, a small feature or a slice of a bigger one. A ticket too big to review well in one pass gets split on the scope call, before the price is fixed, so you know the PR count up front.
How do payment terms work, who invoices, and who pays for LLM and API usage?
Every price is fixed before work starts. Prices exclude VAT; VAT is added where it applies. Invoiced in EUR by my Romanian-registered business (PFA); USD on request. The Architecture Session is paid up front. Sprints are 50% to book the slot and 50% on handoff, the Reviewed PR is paid after delivery, and retainers bill monthly. LLM/API usage is included in the fixed price for sprints and reviewed PRs. Agent Fleet Setup runs on your team's own Claude Code licences.
Why not a contractor by the day, an agency, more engineers, or an AI review bot?
Day-to-day capacity is the retainer, from €4,000/month. An agency sells a team's hours with a markup, and new hires take months to ramp. An AI review bot is cheap and worth running, but it can't be held accountable for a merge. A sprint sells reviewed PRs at a price fixed up front, and the first one lands within days. To test the working relationship first, start with one Reviewed PR.
Tell me the backlog. I'll tell you the number.
A 30-minute call. The scope and a fixed price follow in writing within one working day.