Technology Executive - Fintech, Payments & Digital Assets
I build the systemsambitious companies run on.
Six years as a full-time Chief Technology Officer in three different businesses - regulated lending, crypto media, payments - doing the same work each time: restore a system's fundamentals, then give it room to scale.
01 - The Throughline
Fifteen years, five companies, one recurring position
Fifteen years in technology. Six of them as a full-time Chief Technology Officer, three times over: first at Webbankir, a regulated online lender, then for four years at Cointelegraph, a crypto media platform read by millions of people every month, and now at XPlace, a platform that lets people borrow against crypto and spend it without selling it. Three different businesses with very little in common.
The work was the same each time. A system already existed, built by people who did what the moment required. It was now carrying more weight than it was designed for, and it had to hold without stopping the business running on top of it.
That's the throughline. Not a single technology, not a single industry. The recurring position of being handed a system at the exact point where informal coordination stops working, and being trusted to fix what's actually broken instead of whatever is easiest to fix.
02 - How Systems Fail
Rarely for lack of effort
A system rarely fails because nobody is trying hard enough. It fails because effort has been going into the wrong place for long enough that the wrong place starts to look permanent - a monolith nobody wants to touch, a compliance check that depends on one person remembering, a vendor bill nobody can explain.
The fix is never more effort applied where it already is. At Webbankir it meant replacing a distributed monolith, services all writing to the same database and failing constantly, with services that had real ownership boundaries and talked to each other only through events and APIs. It came to about sixty services in production by the time I was CTO. At Cointelegraph it meant treating a turnaround not as a crisis to survive but as the moment to rebuild: stop a replatforming effort that had gone nowhere for two years, consolidate a scattered service estate into one backend, and only then move delivery in-house. Crises are where an organisation grows fastest, if the fundamentals get fixed first.
Different companies, different failures, the same method underneath: find what the system is actually built on, rebuild that part first, and let the rest follow.
03 - Two Domains, One Discipline
Fintech asks for trust. Media asks for speed.
Running engineering for both, back to back, meant unlearning the previous domain's instincts on the first day of the next one.
A lending platform lives on trust. Every decision has to be defensible, auditable, and deliberate enough to be right - where deliberateness, left unchecked, is exactly what stalls a company that needs to grow.
A platform lives on attention. It has to publish constantly, move fast, and answer to a search algorithm that changes without warning - where speed, treated carelessly, is exactly what invites trouble.
What carried across wasn't a playbook - fintech and media don't share one. It was the discipline underneath both: define what the system is actually optimising for, build the organisation and the platform to match, and never confuse one domain's definition of "done" with the other's.
04 - AI as Infrastructure
It changed how fast I can act, not what I believe
At Cointelegraph, LLM-based and agent tooling - including Claude-based systems - became part of the operating layer, not a side project. AI went inside the editorial workflow rather than sitting beside it: generated SEO titles, descriptions and cover images, AI-translated language editions, and policy compliance checks that blocked non-compliant articles from publishing. The point wasn't novelty. A check that stops an article before it goes live is a control built into the work, not a step someone has to remember.
The interesting question was never which model to use. It was which parts of a workflow could finally be rebuilt now that checking, auditing, and validating no longer required a person to be the bottleneck - and which parts still needed one. Getting that boundary right is most of the job.
Keeping current with the tooling itself - new models, agent frameworks, the systems now connecting them to real infrastructure through protocols like MCP - is a standing practice, not a research interest on the side. The organisations that benefit are the ones where evaluating a new tool and deciding whether it belongs in production has become a routine engineering skill, not an occasional executive initiative.
05 - The Record
Selected experience
Chief Technology Officer - XPlace - Dubai, remote - February 2026 to present
A non-custodial platform that lets people borrow against their crypto and spend it with a payment card instead of selling it. The user keeps their keys, the collateral sits in a lending protocol, and the card spends against it. Built on Solana, as an orchestration layer over regulated partners.
Owns architecture, delivery, security, infrastructure, budget, the partner and vendor stack and much of the product definition, working day to day with the CEO and a team of eight engineers across backend, frontend, on-chain, QA and DevOps. Runs the payments surface end to end: card issuing, virtual accounts for on- and off-ramp in the US, EU, UK and Brazil, swaps and multichain transfers. Card transaction success stays close to 100% against monthly card volume growing 30%+. Built transaction monitoring from scratch, tracking the success rate of every operation type (deposits, withdrawals, transfers to virtual accounts, card authorisations, cashback, swaps) with alerting that surfaces failures before customers report them. Works with payment partners on compliance and anti-fraud: transfer validation, monitoring, the controls partners require. Built the B2C support function. AWS and Kubernetes, with a fault-tolerant database.
Fractional CTO - HABER - part-time, remote - November 2024 to present
A registered professional collections business that moved from an agency model to buying consumer debt portfolios onto its own balance sheet. Joined at that transition to build the technology the new model needed: a debtor portal where people see the debt and settle it, through refinancing, a discounted one-off settlement or a payment plan.
Made the buy-or-build call on evidence: piloted an off-the-shelf collections product, concluded it would not carry the business, and built an in-house CRM instead - in-app telephony, automated case allocation and task generation, workflows for lawyers, finance and payments, court filing and enforcement tracking. Underneath it sits the money and reporting layer: three banks for automatic reconciliation of court-case receipts, a payment provider for debtor payments, credit bureau reporting, a CDP over Kafka, cloud infrastructure with monitoring, and BI dashboards on collection performance. Designed and led the work personally at the start, then hired the technical manager who now runs it day to day, with a team of around ten across development, QA, product, analytics, security and DevOps.
Chief Technology Officer - Cointelegraph - remote - May 2022 to April 2026
Joined into a turnaround. The previous technology leadership had been let go, delivery ran through outstaff agencies on several continents, two years of work on a new editorial system had never shipped, and nobody could say what the company was paying for. At its peak the organisation was 40 to 45 people across web, CMS, mobile, QA, DevOps, product design, SEO and special projects, with managers reporting in, behind a platform read by 12 to 13 million people a month in 12 language editions, at 99.9% availability, on infrastructure across the US, Europe and Asia.
Cut engineering cost by 70% and infrastructure cost by 30% in the first year, moving delivery from outsourced agencies to an in-house organisation that was hired and structured in their place, and unwinding vendor contracts that took hard negotiation and, in places, litigation. Stopped the failed replatforming effort and rebuilt instead: consolidated a scattered service estate into one backend, migrated the reader portal from Vue 2 to Vue 3 and made it load several times faster than the nearest competitor, moved hosting to Kubernetes with Cloudflare geo-balancing and multi-region failover.
Built a new CMS from scratch with real-time collaborative editing, then put AI inside the editorial workflow. Grew QA from two people to seven, with automated coverage of about 80% on the backend and 60% on the frontend, wired into CI alongside linting, unit tests and dependency vulnerability scanning. Moved delivery to two-week sprints, shipping features as they became ready rather than assembling release builds. In the second half took on product: product, product design and SEO reported in, with the roadmap set together with the owner. Owned the technology budget and payroll, with cost reviews every two to three months.
Chief Technology Officer - Webbankir - Online Lending, Moscow - March 2020 to May 2022
Promoted to CTO when the previous CTO left at the start of the pandemic, having run engineering in practice for a year. Owned platform, infrastructure, data, support, budget and vendors, reporting to the CEO. Six months into the role joined the company board alongside the CEO, CFO, CMO and the heads of risk and business development, meeting weekly on business and financial results and setting direction on approval rates, collections and product priorities. The engineering organisation was 40 to 45 people across backend, frontend, mobile, DevOps, data, QA and the credit conveyor, with about 60 services in production and roughly 150,000 loans disbursed every month.
Redesigned the application flow around the cost of each check: free verifications first, cheap paid ones next, the most expensive third-party checks only for applicants who had already qualified. Borrower verification spend fell by about 50% with no loss of decision quality. Rebuilt the loan decisioning pipeline on a new system, several times faster and changeable by the business without an engineering project each time. When new data-protection legislation required certified hosting, migrated to a compliant cloud: the incumbent provider quoted a 30% budget increase, so the vendor changed instead, and the platform ended up with more capacity for roughly half of the pre-migration spend. Event-driven service boundaries let features ship per service instead of in batched releases with full regression. Built data and observability on Prometheus and Grafana, with dashboards on loan volume, quality and status progression, the collections CRM moved onto live platform APIs, ETL, and company-wide BI on Power BI and Redshift. Owned budget, engineering compensation and bonuses, weekly one-to-ones and a performance review cycle. Stack: PHP/Symfony, PostgreSQL, MongoDB, RabbitMQ, HashiCorp Nomad, GitLab, Jenkins.
Head of Development - Webbankir - Moscow - August 2018 to March 2020
Joined as a senior engineer and was promoted within months, after closing a backlog of integrations the company had been stuck on: government identity services, payment processing, address normalisation across the customer base. Also ran the company's second engineering office, opened to make hiring easier.
Inherited a distributed monolith, services all writing to the same database, competing and failing constantly, and rebuilt it into services with real ownership boundaries and event-driven, API-only interaction. Rebuilt the backend team, expanded the frontend team and rewrote the frontend on the same principles. Took over infrastructure after repeated outages: HashiCorp Nomad, fault-tolerant hosting, CI, automated testing, health checks and deployment, rebuilt from scratch. Moved finance, legal and marketing from ad-hoc requests to planned work through Jira. Became the de facto owner of engineering across the company while the formal CTO was largely absent.
Head of Web Development - Comindware - Low-code BPM, Moscow - August 2015 to August 2018
A low-code BPM platform sold to enterprise customers in North America, the DACH region and CIS: insurance, healthcare and public-sector organisations, including a United Nations agency. Led the engineering team behind the systems the company sold through: the product site and landing pages, the backend serving a two-way Salesforce integration, email and SMS notification pipelines, and automated provisioning of trial environments on AWS.
Proposed splitting the team lead role, moved the previous lead into a full-time DevOps position and took over the team, growing it to seven engineers across backend and frontend plus a dedicated QA, so the team no longer depended on the product team's testing queue. Rewrote backend and frontend almost from scratch. Designed the two-way Salesforce synchronisation, including record priority and conflict resolution, inside the API quota of the existing tier. Introduced weekly releases where there had been no release discipline. Joined in 2011 as a frontend developer and grew through senior frontend and full-stack roles into leading the team. Stack: PHP/Symfony, Redis, AWS, Jenkins, WordPress, jQuery moving to early React.
06 - What I Believe
Five convictions, not a values slide
A system's fundamentals matter more than its features.
Features get noticed. Fundamentals decide whether the features survive contact with real load, real regulation, and real time zones.
Management is a discipline, not a promotion.
Fifteen years into the work, still studying it formally - currently under Vladimir Tarasov at the Tallinn School of Management - because instinct plateaus and method doesn't.
AI is infrastructure until proven otherwise.
It belongs wherever it removes a bottleneck a person shouldn't have to be, and nowhere else. That's a boundary worth checking often.
Teams across time zones still share one outcome.
Distributed engineering only works when accountability is structural - a performance review cycle, regular cost reviews, weekly one-to-ones - not a standing meeting.
The recovery matters more than the crisis.
Anyone can describe what broke. The measure is whether traffic, trust, or reliability actually comes back - and stays back.
07 - Where I'm Useful Now
Three conversations worth having
A platform or a team that has outgrown the architecture - or the org chart - it started with, and needs both rebuilt without a pause in the business.
An engineering organisation deciding what AI actually belongs in its workflow, past the pilot stage, at the level of production systems and everyday operating rhythm.
A board or an investor trying to evaluate whether a technology function is actually built to scale, or is currently just coping.
If one of those is the situation, that's the conversation worth starting.
