All projectsReal-time backend

Real-Time Market Data Service with Multi-Source Aggregation

A web service that pulls live market data from three independent exchange APIs, cross-checks the responses, and returns a single aggregated view through a REST API.

Role
Backend Engineer
Evidence
3 independent sources · degrades instead of failing · public demo
Status
Public demo, private core
PythonFastAPIDockerREST APIExchange APIs
Open public product

The problem

A product that trusts a single data provider inherits every outage, delay and malformed response from that provider. Trusting all of them blindly is not better: prices disagree, response shapes differ, and aggressive polling reaches rate limits quickly.

What I built

A service that pulls live market data from three independent exchange APIs, cross-checks the responses, and returns a single aggregated view through a REST API.

  • Degrades instead of failing. Built for reliability rather than features: if one source is slow or down, the service still responds from the others.
  • Caching with rate limits in mind. Data is cached to keep latency low and stay within provider rate limits.
  • Validation at the boundary. Every response passes through validation before it reaches the client, so broken upstream data does not leak through.
  • Deliberately lightweight. Three dependencies, containerized, deployed on cloud infrastructure and running 24/7.

Public interface

The live deployment reads three venues — Bybit, Binance and OKX — on a 60-second refresh and publishes an algorithmic read of the market: a per-asset verdict for BTC, ETH, SOL, XRP and HYPE, a four-day movement map, and ranked liquidity magnets above and below price. The interface is in Russian, and it carries its own disclaimer: it is not investment advice.

Stack

Python, FastAPI, Docker, REST API, exchange APIs.

PUBLIC DEMO / PRIVATE CORE

The public interface can be inspected directly. Provider configuration, credentials and the scoring logic behind the published view remain private.

Ready when the problem is real

Where does manual work become expensive for you?

Describe the process, the failure you cannot afford and the result you need. We will propose the smallest system that solves it reliably.

Discuss a similar system