Ivan MakukhinDelivery Director & AI Transformation

Ivan Makukhin at the window of his office in Iași

Programmes that were supposed to be impossible.

I run large delivery programmes in regulated industries, and I fix the ones that have stopped moving. Fifteen years, fifteen markets, budgets to $32M. Lately the work has been about getting AI out of the pilot stage and into production, where it has to survive an audit.

Next availability: about a month out

Iași, Romania Remote across the EU linkedin.com/in/ivanmakukhin
300+
people in the largest programme I have run
$32M
largest programme budget delivered
10M
customers on a platform rebuilt in nine months
100+
manufacturing sites in scope on a live programme

What I do

Four things, and they are usually the same engagement seen from different sides.

Run the train

Release Train Engineer and programme lead for up to 9 teams and 14 squads. PI planning, cadence, dependencies and risk — and governance that leadership can actually act on.

See cases 01 and 02 ↓

Get AI into production

From scattered pilots to something that runs and gets used, with adoption and business effect measured rather than assumed.

See case 03 ↓

AI governance in regulated environments

Data boundaries, traffic inspection, DLP, access control, human-in-the-loop. The part most consultancies are still staffing up for.

See cases 03 and 04 ↓

Recover programmes that lost the plan

Find the real defects, change the approach, deliver clean. I have rolled back a failed implementation and replaced the provider mid-flight.

See case 07 ↓

Selected work

The problem first, because that is what you are actually buying.

01RTERetail banking

Rebuild a bank's platform in nine months without letting compliance slip

A Western European retail bank had to replace the mobile and web platform serving ten million customers. Like-for-like scope, a fully redesigned experience, new SEPA payments, tokenisation and WCAG AA accessibility. The regulatory date could not move.

What I did Took the Release Train Engineer seat across seven teams and fourteen squads, three delivery managers and seven scrum masters, around three hundred people at peak. Built the cadence so regulatory review sat inside the release cycle instead of behind it, and ran one dependency and risk picture that the steering committee actually used to decide. Presale through execution.

9 months10M customers300+ peopleSpring Boot, MuleSoft
02Agile Coach → RTEPharmacy & healthcare retail

Grow two teams into a release train when nobody has run one before

A global pharmacy & healthcare retailer had two, at times three, teams on a consumer healthcare platform and needed many more. Nobody in the building had scaled a train, and every screen had to meet WCAG AA and HIPAA.

What I did Came in as Agile Coach and took the RTE seat as the train grew. Took the full W3C front-end programme myself, turned it into team training on WCAG AA and HIPAA and a shared bug knowledge base, then grew scrum masters and handed teams to them one at a time. Figure it out yourself, package it as training, raise the teams: that is the method, and this is where it is visible end to end.

2–3 → 9 teams90 people on the train−50%+ known defect backlog
03AI Delivery LeadAgri-science manufacturing

Take AI past the pilot stage at a manufacturer with a hundred plants

A global agri-science manufacturer had AI experiments running and no route to production. At the same time confidential data was going into public AI tools, because policy said not to and nothing stopped it.

What I did Took the AI Delivery Lead seat and owned the rollout as its product owner. Delivered the first enterprise-wide AI service — an assistant with a policy and knowledge-base module, a testing model and a phased go-live. Put AI traffic inspection across the estate so the data boundary became something enforced rather than written down. Built the AI-maturity measure now applied across delivery programmes.

First production AI service in the groupGovernance enforced at the network layer
04Transformation Lead · Zero TrustAgri-science manufacturing

Move forty thousand users to zero trust while rebuilding the network under them

The same manufacturer needed to retire its legacy VPN across a global estate, and separately to modernise the network across every plant — two programmes touching the same sites, the same change windows and the same people.

What I did Transformation Lead and Product Owner for Zero Trust on the Zscaler rollout: segmentation, access policy, rollout waves and L3 support ownership. Before that, as Network Programme Lead, ran the $32M network programme across the plant estate as one planning, tracking and governance system covering plant, OT and IT scope. Built the first secure IoT perimeter for manufacturing devices.

40,000+ users−60% security incidents per enabled user99.9% uptime100+ sites
05Programme ManagerVeterinary pharma

Replace a CRM in a dozen-plus markets, one country at a time

A European veterinary pharma client was replacing its CRM across 12–15 markets. Every market ran sales its own way, and the legacy platform had grown around each of them.

What I did Programme Manager on a full transformation cycle rather than a CRM project: mapped the business process and the pain market by market, replaced the legacy platform, rewrote the process in each country before its go-live. Personally delivered seven market rollouts in about ten months, with two more in flight at handover. Nexus for scaling, a mix of fixed price and time and materials.

7 markets delivered personally~10 months95% adoption95% CSAT
06Director, Supply Chain & R&D ProjectsMine safety equipment

Ship certified hardware without pretending the release plan is software

A mine safety equipment manufacturer builds products in a certified category. Validation testing takes what it takes, tooling has a lead time, and no amount of planning pressure compresses either.

What I did As Director of Supply Chain & R&D Projects, ran twenty-plus development projects from concept through certification testing into series production. Built procurement from one person to nine and logistics from nothing, including the buy side of a greenfield plant build in Turkey — long-lead equipment, supplier qualification, commissioning sequence. The programme plan was built around the validation cycle rather than against it.

20+ products to series−20% supply chain cost−30% cycle time−15% defects
07Transformation Lead · Zero TrustAgri-science manufacturing

Restart a data-loss prevention rollout that had already failed once

The implementation was in flight and not working. The usual answer is to push harder and hope the next milestone looks better.

What I did Same seat as the zero-trust work, Transformation Lead and Product Owner: this was part of the security track, not a separate hire. Found the defects that actually mattered, rolled the deployment back, changed the provider and delivered it clean. Saying that a programme has to stop is the expensive part; doing it early is the cheap one.

Rolled backRe-scopedDelivered clean

Also. For a national statistics authority, as transformation PM: put the country's first three statistical surveys onto a modern data platform and took manual data handling out of the process.

Client names, sanitised detail and references on request.

How I work

Four shapes. The first one is most of the work, and the last one is the cheapest way to find out whether you need the first.

Ivan Makukhin at a wall of planning notes with two colleagues, Iași
Planning session, Iași
01Interim delivery leadership6–18 months · full time or close to it · time and materials, invoiced monthly

A running programme needs someone to hold the seat, now.

Details ↓
02Fractional delivery leadership1–2 days a week · rolling, 3 months minimum · monthly retainer

You need the seniority and not the headcount.

Details ↓
03Assessment, launch or reset2–6 weeks · fixed scope, fixed price · priced above the long-engagement day rate

Something has to be stopped, restarted or launched properly.

Details ↓
04AdvisoryHourly · single call or standing slot

You mostly know the answer and want it argued with.

Details ↓

How it starts

  1. 1A call. Thirty minutes, free.

    You describe what is stuck. I say on the call whether it is mine, and if it is not, who to talk to instead.

  2. 2A short paid look.

    For a long role I do not sign blind. I go in, see it live and hand you a written recommendation: what to fix, in what order, in what shape. Priced on its own and small next to the role. The document is yours to take to someone else. This is the advisory shape above, sold as a step.

  3. 3The work itself.

    In the shape step two proposed, on the terms below.

Already know what you need, and it is a plain time-and-materials seat? We skip step two, agree terms and start. That is how most agency work goes.

01

Interim delivery leadership

RTE, programme director or delivery lead on a running programme. I hold the seat, not a slide deck: cadence, dependencies, risk, the steering conversation, and whatever else the role turns out to include.

Six months is one full cycle — long enough to change how a programme runs and watch the change survive contact with a quarter close. Twelve to eighteen is where it keeps running after I stop being the reason it does. My own assignments have averaged nine to twelve months, which is about right for the work.

02

Fractional delivery leadership

One or two fixed days a week, ongoing. For a company that needs the seniority and does not need the headcount, or is not ready to hire for it yet.

Same days every week, so teams can plan around them and I am not a calendar negotiation. Monthly retainer rather than a timesheet — the point is that the days are reserved.

03

Assessment, launch or reset

A delivery assessment with findings someone can act on. Launching an ART and designing the first PI planning, then facilitating it. Or taking a programme that has to be stopped, stopping it properly and restarting it on something that works.

Short work costs more per day than long work. There is no utilisation behind it and almost all of it is ramp-up — I am paying the cost of learning your environment across two weeks instead of twelve months. I would rather say that plainly than bury it. Very short pieces I take case by case — the ramp-up has to be worth it for both of us.

04

Advisory

By the hour, no ramp-up. A plan you want argued with, a framework choice, a vendor decision, a delivery org you are about to restructure, a rate you are about to agree.

Useful when you mostly know the answer and want someone who has done it to tell you where it breaks. It is also step two of how work starts: the short paid look before a long role, priced on its own.

Contract

B2B, company to company. Invoice from an EU entity with a VAT number, bank transfer. Your paperwork, your framework agreement — I have signed enough of them.

Where

Remote, working to your hours. On site for planning events, the first weeks of an engagement, and whenever the room matters more than the call.

Rates

Quoted against your market and the shape of the work, in EUR, GBP or CHF. Tell me the budget you are working to and I will tell you straight away whether it fits.

What I turn down. Single days and one-off favours dressed as engagements — those are advisory calls, and they are cheaper. Staff augmentation priced as a body on a rate card. And any seat where I would be the fifth hand-off between you and the work, because nothing I do from there is worth what you would be paying for it.

Credentials

Enough paper to pick the framework from the diagnosis, not from habit.

PMP
Project Management Institute
SAFe RTE
Release Train Engineer
CSP-SM
Certified Scrum Professional
MBA
University of the People

Certified in every major scaling school — SAFe, Nexus, Scrum@Scale, Kanban. That is deliberate. It means I can pick the framework from the diagnosis rather than from the one I happen to have been trained in.

More than twenty-five in all, across PMI, Scaled Agile, Scrum.org, Scrum Alliance, Kanban University, Microsoft and GitHub. All of them verifiable at credly.com/users/ivan-makukhin.

English · Ukrainian · RussianWorking languages. I run programmes in all three.

Work with me

Tell me what is stuck and when it has to be fixed.

ivan@agileai.tech
+40 758 937 705 LinkedIn Iași, Romania · EET, UTC+2

I will tell you within a day whether I am the right person for it — and if I am not, usually who is. If you already know the shape you want, say which of the four it is and we can skip a step.

Ivan Makukhin in conversation at his desk
Founder · AGILEAI TECH S.R.L.