This is the story of a senior ML engineer placement at a Fortune 500, publicly-listed US fintech that needed PhD-grade depth in pricing optimization and production-grade ML systems
This engagement was led by Kamen Bochev, founder of Perpetum, in spring 2025 while operating as Chief Delivery Officer at a partner staffing firm. The recruiter who led the search is now part of Perpetum’s core team. The methodology, the operating standards, and the people are the same – Perpetum is the continuation, sharpened.
The Ask
A Fortune 500, publicly-listed US fintech needed a Senior Machine Learning Engineer with PhD-grade depth in pricing optimization and production-grade ML systems. The brief was specific: not a research scientist, not a generalist data scientist – an engineer who could build pricing models that directly affected product margin, deploy them to production, and operate as a peer inside a large cross-functional distributed team.
The client’s European headquarters were in the UK, where a senior engineer of this profile would have cost approximately three times what an equivalent hire would cost in Eastern Europe. They had already done that math internally and made the strategic decision to source the role from the region – drawn by the quality-to-cost balance that Eastern European engineering talent pools offer for senior technical roles. What they didn’t yet have was a partner who could actually deliver the profile they needed within their budget framework.
The combination they were looking for – academic depth, hands-on production engineering experience, and the maturity to operate inside a complex enterprise environment – is rare anywhere in the world, and effectively unavailable in their UK market at a defensible cost. Eastern Europe was the answer to the cost question. Finding the right specific human inside Eastern Europe was a different problem.
They came to us with a tight budget framework, a long technical bar, and a need to move quickly. The director championing the role was actively trying to modernize how the organization deployed ML to production, and every week without the right hire was a week of stalled progress.
The Solution
Before sourcing, we sat with the client and worked through the role in detail. What did the team actually need versus what the brief said? Where would this engineer spend the first 90 days? What did “PhD-grade” mean in practice – pure research credentials, or applied depth in pricing and optimization? Getting this right at the front end is what separates a placement from a search.
We then designed a multi-stage technical assessment process tailored to the role. Not a generic algorithms test – a structured evaluation that included a complex applied take-home and a system-design conversation focused on production deployment patterns. The assessment was demanding enough to filter out candidates who looked strong on paper but lacked the production engineering instinct the client actually needed.
From a shortlist of strong candidates we presented two finalists. The client selected one, the technical evaluation went well, and an offer was extended within the agreed budget framework.
The first candidate didn’t close.
Mid-process, he received a competing offer at a higher compensation level than the client’s framework allowed. We escalated, explored whether the budget could flex, and confirmed it could not. Most agencies would have gone back to square one at that point. Because we’d built a deeper bench during the search than the brief strictly required, we had a viable plan B already in process.
Plan B closed. The candidate had competing options of his own, and convincing him to take the engagement required real conversation about the role, the team, the trajectory of the work, and the kind of problem he’d actually be solving. He joined within the original timeline.
The Outcome
The placement worked, but the start was rough. The client’s environment was a complex matrix of large cross-functional teams, established habits, and slow deployment paths – exactly the situation the director had brought the engineer in to help change. The engineer’s job, in practice, wasn’t just to build models. It was to operate as a senior technical voice inside an organization that wasn’t fully ready for the change.
We stayed close to the engagement through the first ninety days. Active account management, regular check-ins with both sides, problem-solving when integration friction came up. Standard work for us, but worth flagging because it’s the part most staffing firms quietly skip after the placement is signed.
The engineer is still with the client. As of late 2025, he’s being promoted into a different team where his skills can be leveraged on faster-shipping initiatives – a recognition by the client that the hire was significantly stronger than the role he was originally hired into.
From the client’s perspective, the engagement closed a skills gap that had been blocking initiatives, accelerated their pricing optimization work, and proved out a model for accessing senior technical talent that they couldn’t find locally at a defensible cost.
From ours, it’s a reminder of why we operate the way we do: deep search, structured assessment, a bench deep enough that plan A failing isn’t a crisis, and active management through the rough early months. The placement isn’t the end of the work. It’s the start of it.
What this engagement demonstrates
- Specialist sourcing capability. Access to PhD-grade engineering talent that local markets often can’t supply at defensible cost.
- Custom assessment design. Technical evaluation tailored to the role, not generic screening.
- Contingency depth. A bench deeper than the brief requires, so plan A failing doesn’t restart the clock.
- Active account management. Engagement support through the first ninety days and beyond, where most placements quietly fail.
When the local market isn’t delivering
If you’re hiring a senior specialist and the local market isn’t producing the right profile, this is the conversation we’re built for. Reach out – we’ll get back to you right away with a clear yes or no on whether we can deliver.