Scaling Real-Time Copy Trading for Millions of Users

Copy trading has become one of the most notable trends in retail investing, promising everyday users access to the strategies of top performers. The real challenge lies in the technology that can mirror trades instantly, securely and at scale.

In the , dub became the first copy-trading platform to deliver a compliant, fully regulated trading experience through its licensed brokerage partners — and much of that success is tied to the work of Andrii Humeniuk, a Ukrainian engineer with deep expertise in building scalable systems across global tech hubs. Joining at the prototype stage, he assisted turn dub into a production-ready platform used by tens of millions of users. In 2025, dub raised a Series A and became a major player in the market.

In this interview with Finance Feeds, Andrii explains how he built dub’s architecture to handle billions of transactions each month, why real-time copy trading is technically complex and how transparency can reshape retail investing.

Andrii, could you tell us how your work on the dub platform began? What responsibilities did you take on when you joined the company, and what was the most challenging aspect at the begin?

I joined dub when the platform was still in an ahead prototype stage. I liked the idea right away and its potential was clear, but the system needed an architecture that could grow into a full-scale real-time trading platform. My first focus was to take that ahead version and turn it into a production-ready system that could handle live trading, portfolio replication and the financial compliance requirements that come with scale.

I led the team that developed dub’s core trading and rebalancing engine, which drives all real-time portfolio synchronization. I introduced an event-driven, state-machine architecture to make sure every transaction is handled with accuracy, fault tolerance and full traceability. This made it possible to mirror creator portfolios across thousands of users with sub-second consistency.

I also worked on shaping the engineering culture and internal processes, from code reviews to CI/CD and observability.

So you were one of the first engineers in the company and essentially transformed a prototype into a market-ready fintech product. In your view, which architectural and product decisions were most significant for bringing the application to market and attracting investor interest?

One of the most significant decisions was to design the platform as an event-driven distributed system instead of a conventional service stack. Every operation, whether a trade, a rebalance or a payout, is represented as a discrete event flowing through Kafka streams. This approach gave us horizontal scalability, full traceability and strong fault isolation across all trading workflows.

We paired this with finite-state machine patterns for orchestrating transactions. Every portfolio action moves through clear stages, with every transition persisted and idempotent. This makes execution deterministic and lets us recover cleanly from failures without losing data.

On the product side, we embedded analytics and engagement features directly into the trading core, which assisted turn dub from a trading app into a real-time social investing ecosystem.

Dub is real-time copy-trading that allows retail users to mirror the trades of hedge funds, financial influencers and even politicians. Which of these key features were the most challenging to implement from an engineering perspective, and what role did you play in their launch?

The most complex and defining feature was the real-time copy trading engine itself – the system that mirrors every trade from a creator’s portfolio to thousands of follower accounts almost instantly. When a creator bought or sold an asset, those actions were propagated reliably, consistently, and in the correct order across the entire network.

The largegest challenge was achieving speed, accuracy and fault tolerance at the identical time. We needed to synchronize live market data, user allocations, and broker confirmations in real time without ever duplicating or losing a transaction. I led the architecture and implementation of this engine, building an event-driven workflow controlled by a finite state model.

To support this, we built a hybrid in-memory and cold storage model combining Redis and PostgreSQL, which let us process billions of events each month while keeping portfolios consistent and auditable. I also introduced dynamic Kafka partition rebalancing and idempotent transaction secureguards so each trade could be executed only once, even under retries or concurrent processing.

Dub is responsible for processing financial transactions and operates under strict financial regulations. What technical strategies did you design to ensure security, compliance, and reliability while keeping the product quick and intuitive for users?

From the begin, I treated security, compliance, and reliability as core parts of the architecture, not as layers added later.

We designed a multi-layered reliability model using deterministic state machines, distributed consistency and strong observability. Each workflow runs through an idempotent finite state process with clear checkpoints, which lets the system recover securely from failures without losing data or duplicating a trade.

For compliance, we built auditable pipelines where every event is immutably logged and traceable.

To maintain performance, we used selective caching and a hybrid storage setup: Redis for live state, PostgreSQL for durable transactions, and S3 for long-term retention.

Your specialization lies in building architecture “from scratch” to enable rapid business growth. What universal principles would you highlight for beginups that aim to scale as rapidly and sustainably?

My approach is to design and build systems that deliver both velocity and control. For me, the core principles are straightforward. The first principle is modularity before scale: begin with a clean architecture and move to microservices only when required.

The second is observability as culture. At dub, every service emits structured metrics, traces and logs. This made it possible to view system behavior in real time, debug incidents in minutes instead of hours and measure exactly how changes affected performance.

The third is automation over heroics. CI/CD pipelines, environment parity and reproducible infrastructure through Terraform or Kubernetes are not luxuries – they are how you protect speed while keeping consistency.

The fourth is designing for failure, not for perfection.

And finally, every technical decision needs to support a clear business outcome.

In your work with dub and your earlier experience at Glovo, you encountered diverse cultural and technological contexts. How did those experiences – in Spain, the U.S., and Ukraine – shape your leadership style and technical decision making today?

Each environment I worked in taught me something diverse about building technology and leading teams. In Ukraine, where I begined my career, the engineering culture is very hands-on and resourceful. Teams solve complex difficultys with limited tools, and that mindset shaped my foundation – deliver no matter the constraints and value technical depth and precision.

At Glovo in Spain, I learned what it means to operate at scale. I led infrastructure initiatives that improved observability and resilience across teams, and that experience taught me how to design for global reliability while still enabling local autonomy.

Moving to the United States and joining dub pushed me to bring both perspectives together and apply them in a quick-moving fintech environment. Today, my leadership style is a blend of those three worlds.

Which technologies or trends in fintech do you find most promising, and where do you view the largegest challenges for yourself in the coming years?

is entering a phase where real-time intelligence and transparency are becoming the new standard. The most promising technologies are the ones that make financial systems more open, explainable, and efficient without sacrificing trust.

I view huge potential in three areas. The first is programmable finance and composable infrastructure. API-driven brokerages and modular financial primitives let companies build trading, payments or lending capabilities almost like Lego blocks, which speeds up innovation but also demands stronger architectural discipline.

The second is AI-powered analytics and personalization that allow systems to understand user behavior, portfolio risk and market context in real time.

The third is blockchain-based settlement and asset tokenization.

For me personally, the challenge is staying ahead of the complexity curve – leading teams that can integrate AI, data streaming and security at scale without losing simplicity.

What is your broader vision for how dub, and fintech in general, can democratize investing and empower a new generation of retail investors?

With dub, we’re building a system where transparency replaces privilege. Anyone can view how top creators – analysts, fund managers or even public figures – allocate their portfolios, learn from their strategies and automatically mirror those moves in real time. What once required millions in assets and a full trading desk can now be done from a phone with a few taps.

Democratization also depends on trust and education. My goal is to make these systems explainable – every trade, rebalance and performance metric should be visible, auditable and simple to understand.

My vision is to make professional-grade investing accessible to everyone, not just people with institutional access or deep financial expertise. I want to keep pushing that evolution – where finance feels as open and collaborative as technology itself, and where anyone, no matter their background, can build wealth through shared intelligence.