Mobile Banking: A Branch in Your Pocket

Maxim starts with the preconditions: LEGO sets at ten to twelve, where the point was not the instructions but the imagination — turning an aeroplane set into a boat and playing with the boat; a Moscow children's computer club, closed now and a rarity then, with Logo worlds, Pascal, Delphi and .NET; and a schoolboy side job as a system administrator that grew out of acquaintances asking for help with Windows. Then came MEPhI, and he is proud to work in his own field. From his third year he moved to a full-time job at Favorit Systems, a small, almost home-made company: information systems on top of Oracle and Delphi, plus .NET and Silverlight, and orders for cash machines, sales outlets and oil and gas firms. Alexander notes that he wrote his own thesis in Delphi, and that twenty years ago there were not many such opportunities around universities — such little firms were the outlet, where a short spell let you touch many technologies.

In 2011 an order arrived from Moscow's information technology department for an app to book doctor's appointments: Maxim and a colleague built the first Android and iOS version cross-platform on Xamarin — a bet on an unpopular technology that paid off, and that version reached production. Having hit the ceiling of his growth, he left for a gaming media company and became a team lead for the first time: a team of two grew to five, and it was terrifying — on days without written code he was shaking. A year and a half passed with no release, as a social network clone kept turning into chats, geo-chats and dating. He started looking around, HR found his public CV and dismissed him in a single day — painful then, something he is grateful for now. Three criteria shaped the next step: mobile development, a team lead role with growing responsibility, and not a startup but a company that had proved it could survive.

So he joined Tinkoff, today's T-Bank, taking about eight people and two apps: the mobile bank and an app for representatives. He calls those processes bad by today's standards but great for their moment: the spec was a text appendix to the design, releases went out roughly once a quarter with a full regression and no QA automation, and work from payments, debit and credit products fell into one common pool. The first challenge arrived on day one: in August Oleg Tinkov announced that a mobile wallet would ship on 1 December, and the date would not move. The concept was two sentences long; Maxim and three colleagues slept in the office from Friday to Sunday, because the founder's address book of three thousand contacts with megabyte-sized photos weighed about three gigabytes. The release happened, and card-to-card transfers, fines and state debts followed.

The 2015 crisis, with the key rate at 17%, turned the strategy around: relying on credit cards alone was no longer possible, a bank for small and medium business appeared, and two years later investments started the same way — with four mobile developers who only became independent around 2020. Maxim frames his first big challenge as gathering a team not to own it but to hand it over, together with the right culture. In 2020 mobile development grew from twenty people to two hundred, and to roughly three hundred with adjacent professions, so process moved to first place, at times imposed by order. Then came the product transformation: instead of one queue to the mobile bank, autonomous teams attached to their own businesses, plus platform teams holding shared quality and the release calendar.

Dozens of teams had to be passed to the CTOs of the new verticals, with every person asked whether the move interested them — Alexander recalls that some people went to the colleague who owned debit products, while with a colleague from origination they worked through the corner cases of such handovers. Two kinds of glue were needed to keep the organisation from scattering. The first was shared base mobile technologies, which appeared in 2019 not from a grand plan but because three large apps already lived on their own, each with its own scripts. The second was the institution of profession leaders: five heads of mobile development gathered the most respected engineers from every project, split them into tracks — hiring and competency matrices, screening, InnerSource — and handed them the company-wide topics, from shared libraries to tech radars, judging them by what their track achieves. They meet weekly, report to a wide audience every two months, and Maxim feared repeating the companies where one brand hides nothing in common in mobile and people do not know their neighbours.

Maxim derives the technology choice from the business model: the company has no branches, the branch sits in the client's pocket, so 90% of the apps have been native from the start until today, and his Xamarin experience never came in useful; cross-platform would have brought speed and savings, but not that bar of quality. Release-free delivery grew in steps: the application-form generator from around 2017 consumed JSON and drew native views, yet it could not carry business logic and irritated engineers with its custom declarative style. The next engine took JavaScript and an in-house HTML-like DSL, so that an outside business — down to a cook baking cakes in Chelyabinsk — could reach the city shelf inside the app while the bank handled payments and support. Alexander adds that cross-platform usually spreads where the app is the working tool of a driver or a courier.