From C++ in Tambov to T-Bank
Denis Pakhomov starts with his biography: he was born in Magdeburg, where his father served as a military officer; the family later moved around the country and settled in Tambov. He took his degree in information security at the Tambov university, which, as he puts it, laid not the foundation but important bricks of his understanding of computer science. He credits his uncle, a programmer-engineer in the Soviet defence industry, as his inspiration: that is why his first language was C++. University also brought Delphi and cryptography with Feistel networks, and by his fifth year Denis was employed by the institute itself, where the team built a university management system covering accounting and timetabling. Then came Voronezh: a startup checking texts for originality, and in parallel the role of Head of Development at a payment aggregator. Notably, he says, AI models were far less developed then, yet text checking was already done well.
That stretch, by Denis's reckoning, began some eleven years ago and ended about six years ago: with nowhere left to grow, he closed his commitments, resigned and moved to Moscow, where he has worked for the last five years. His first Moscow employer was Sber, where he took into production a system that calculates deals for corporate clients: a large company chooses where to keep its billions, and the bank works out its terms on the minimum and average balances. A second project, on foreign trade, was launched too, but after 2022, in his words, it went no further. He left with the formula that the cover did not match the contents of the book: T-Bank already felt like a different engineering culture during the interviews, and with every section — language, management, system design and the final one with the team, run by Kirill Kraynev — his wish to work there only grew.
MPM, the feed and the first team
His first task at the bank was ML: how to recommend offers to a client when communication slots are scarce and the bank wants to show a lot. The slots are banners in the mobile app, passive channels; on the business side the task was driven by Grisha, and it reached an MVP of an offer prioritisation system, after which the analysts worked out which model performed. Denis had no team of his own then — people from Kirill's organisation, between backlogs, helped out — and Alexander recalls later reading the project's RFCs and seeing Denis's trace in the documentation space. Alexander describes that kind of role as a high-grade individual contributor: leadership without direct management. The next task was already about management and product: the sponsor left the bank, the project stalled, and Denis was handed MPM, a module inside the campaign engine.
MPM answers which promotion a client is in and what is left to do before the reward; the master system takes load off support and fixes the experience of someone who meets the conditions and cannot tell whether the card will arrive. His first team built from scratch appeared here — around eight people, if memory serves him, a cosy one — and he took the product to MVP. In 2023 Kirill entrusted him with the feed on the main screen, now called the means of delivery: the team, which now numbers forty engineers including QA and mobile, became cross-functional then, with shared goals over both short and long distances. Nobody, as Denis sees it, was left demotivated, and the business found it clearer how to work with the team. The feed SDK has matured to the point where it is simply embedded into a new app, and video, he answers Alexander, shipped in the last two releases.
One platform, metrics and the department
The focus of 2023 was already social: Denis took on the public profile in the yellow app for the first time — the old social, launched to a small sample of employees with a gentleman's set of a profile, following, your own feed and a for-you feed. During development it turned out that other teams in other business lines were building almost the same mechanics; Alexander calls such products twin brothers raised in different families: identical from above, they differ in feature priorities or in a chronological feed versus a recommended one. Assembling one team and one platform took, by Denis's estimate, a year or even two: business lines have different short-term goals, hard to pack into a single plan. The first stage was several months of architecture work and harmonising requirements, to see where the overlaps were and where to start without investing in unicorns. Then the master systems were chosen — the most mature services, readiest to extend — and the rest dropped: taking that decision was hard, and so was explaining to IT leaders why their service was not the base. They borrowed the process from the neighbours — Alexander's space and the investments team — compiled something not too cumbersome, set up regular architecture committees with ADRs, and shaped the team from the architecture: Denis calls it the inverse Conway manoeuvre.
Product positioning came down to a single client profile running through the ecosystem and a feed that respects the context of its surface: in the black app that means investments, in the yellow one travel and lifestyle are added, while the media projects were postponed — you cannot eat the whole pie at once. The main business metric now is time spent, with author-side and client-side metrics beneath it: the volume of quality and verified content, plus engagement and retention; monetisation is the next step. Verifiability, Alexander notes, sets this apart from other social networks — you can see that a person really does trade shares — and Denis adds that a transaction can be attached to a post. Today the department is 140 people, 14 teams and five directions: Core, product integrations, delivery, web development, and stories with moderation. The platform carries two products: Pulse in the black and yellow apps, and Link, the internal social network for tens of thousands of employees inside the HR services. Denis splits his time roughly like this: 25% on goal-setting with the product people, the same on org design and efficiency work, about 10% on people management; he no longer reads every ADR, stepping in only when a question heats up.
What to take away
- 01Big players such as Ozon reach an ecosystem from the marketplace side and buy banking; here it is the reverse: a bank with loyal clients is adding the missing social layer next to shopping and travel.
- 02Merging near-identical products starts with choosing master systems: some services become the base, others are discarded, and one of the hard parts is explaining that to the people whose solution was dropped.
- 03Denis's direct reports are already at the limit: four direction heads, two platform architects, a technical leader and a project manager — at that span of control it is hard to stay fully present in every one-to-one.
- 04Advice for growing engineers: invest in your own learning, do not fear uncertainty, and pump up your weakest area, usually management; the books Denis names are Drucker and Adizes.