Plata builds large data warehouse team
Three and a half years after Ivan Shchukin arrived at Plata, the Mexican fintech lacked a data warehouse. Now the firm runs a data team of over 50 staff, processes information in real time, and aims to make business-intelligence tools redundant.
Plata’s mission is to simplify and broaden access to financial services across Latin America, aiming to expand beyond Mexico. They offer a range of financial products, including credit and debit cards, cashback programs, investment options, and services tailored for small businesses.
Ivan’s path to fintech was unconventional, starting as an electrical engineer and gaining experience in project management, software development, and data analysis before being recruited by Plata’s Chief Technology Officer.
Building a Data Warehouse
Ivan Shchukin, Head of Data Warehouse at Plata, recalls the moment he was approached by the CTO. “Andrey Shelekhin, our CTO, said: ‘Ivan, you have an amazing skill set for our head of data warehouse role. Come to Plata – we need to create the best product.'” Ivan accepted and set about building something that he believes remains relatively rare in the industry.
Building Data Warehouse
From the start, Ivan prioritized two key principles for the data warehouse: full data collection with full version history and rapid updates within one hour, significantly faster than the typical 24-hour lag.
Ivan explains the thinking: “We want to collect all data in the company with full versioning. Everyone wants this, but most players in the market are afraid of it. I had no limitations and no fears.” The one-hour latency target was not merely a technical preference but a business decision, creating a measurable competitive advantage in retail banking and fintech.
Related Post: Sustainable bond market shows uneven regional growth
Expansion and Growth
The data warehouse department has grown from a team of eight when Ivan arrived to more than 50 people organized across roughly 10 specialist teams. Managing that expansion has been one of the more demanding aspects of the role. Ivan frankly explains the learning curve involved in leading a department of that size.
Alongside managing growth, the team made an early structural decision that Ivan regards as significant: hiring analysts into a discipline called analytics engineering rather than the more established field of data engineering. Analytics engineering is a relatively recent specialism that sits between data engineering and data analysis, focusing on transforming and modeling raw data so that it is ready for business use.
Ivan reveals: “We were among the first to set up analysts with the title of analytics engineer. It gave us the right processes, the right hiring approach, and the right path for growth.” That hiring philosophy now informs Plata’s expansion plans, with Ivan envisioning hiring a small founding team of analytics engineers in each new market, supported by the automation and tooling the department has already built.
Expansion Plans Unfold
The company’s growth and expansion into new countries will rely heavily on its data capabilities, and Ivan’s team is working to develop a single window for data, an internal integrated development environment that will provide a centralized workspace for writing and running code.
Currently, Plata’s staff use around 10 different tools to perform data-related tasks, and consolidating these into one AI-assisted environment is among Ivan’s primary objectives for the year ahead. Ivan continues: “We want just one window for any task – whether that’s a dashboard, a report, or a shadow process – for any user across the company.”
Ivan’s vision for the future of data analysis at Plata involves AI-generated visualizations, with analysts using AI tools to generate bespoke visualizations on demand, written in a Python framework called Streamlit. The plan is not without its challenges, and Ivan is candid about the limitations of simply pointing an AI model at a database and expecting accurate answers.
Read Also: UK Loses $13 Billion in Hidden Fees, Wise Reports
The quality of any AI-generated output depends entirely on how well the underlying data has been described, and metadata must be thorough and accurate before AI can interpret it reliably. Ivan explains: “People say you can integrate an LLM with any database and it becomes a simple computer you can question about anything. But that’s not our reality. You need to describe your data first, and that’s where we are working now.”
Scaling with Partners
Plata relies heavily on Snowflake, a cloud-based data platform, for its scalability and efficiency. Ivan highlights, ‘Snowflake has eliminated many challenges for us. We can expand into new markets or domains in minutes, not months, which is remarkable.’
Ivan goes on to frame the importance of partnerships in terms that go beyond technical convenience, stating that a good technology partner should share the same values and long-term direction as the company it serves. He is blunt about the stakes involved in getting that relationship wrong, emphasizing the importance of alignment and shared goals.
For the next 12 months, Ivan has three clear priorities: geographic expansion of the data warehouse into at least one new country, the launch of the internal IDE, and a significant increase in the proportion of staff using AI tools on a daily basis. Currently, around 60% of the data warehouse team incorporates AI into their daily work, and Ivan wants that figure at 100%, and soon.
Ivan concludes that AI adoption is no longer optional, stating: “AI is like a calculator in 1980. You just need to use it to be relevant; it’s not optional any more.” The company’s data warehouse now underpins approximately 80-90% of its data processes, all running with automated quality checks and alerts, and Ivan’s team is working to further develop and refine this capability.
