Three of our own products, picked because they cover different ground: algorithmic trading, a sports mobile app with an admin panel, and a geospatial analytics platform. Each runs on the same infrastructure we offer clients.
Continuous operation
24/7
Continuous operation
Decision latency
Sub-second
Decision latency
Historical data
6 years
Historical data
Web · Mobile · Data backend
Algorithmic trading platform
A system that runs trading strategies on crypto markets: it collects market data, executes strategies, holds positions and reports results. A web admin panel for configuration and oversight, a mobile app for following positions.
What we built
Live order book over WebSocket: cumulative depth, spread in basis points and bid/ask imbalance
Continuous ingestion of spot and futures candles, order book and funding rates
News feed from public sources, plotted against price day by day, with story counts and sentiment spread
Versioned strategy engine: every change stays reproducible
Backtesting over historical data, with patterns marked only from bars that had already closed
Idempotency on the execution path — a retry never becomes a duplicate order
Stack
TypeScript
Node.js
PostgreSQL
React
React Native
WebSocket
Admin screens
15
Admin screens
Platforms, one codebase
2
Platforms, one codebase
Activity source
Strava
Activity source
Mobile app · Admin panel
Heroes of Terra
A sports app built as an RPG: real runs, rides and swims turn into damage against a raid boss. Players organise into guilds, pick a role, and take part in time-limited events.
What we built
iOS and Android from a single codebase, on Expo
Strava integration: real activities become in-game actions
Guilds, raids, quests, spells, buffs and achievements
Push notifications and an offline mode for areas without signal
A 15-section admin panel for balancing the game
Template editor for events, so a change does not need a deploy
Stack
React Native
Expo
Python
FastAPI
PostgreSQL
React
Business types
46
Business types
Hexagon resolution
~170 m
Hexagon resolution
Score, with reasons
0–100
Score, with reasons
Web platform · Geospatial analytics
Agency — site selection intelligence
It answers one question: where is a given type of business worth opening. It splits a city into ~170 m hexagons and scores each from 0 to 100 on what is actually nearby — universities, offices, hospitals, transit stops, apartment blocks — minus the competitors already trading.
What we built
46 business types, each with its own demand model: what draws customers, and from how far
A 0–100 score per hexagon, placed in context — "in the top 7% of this city for a bakery"
"What drives it": each factor's contribution as a percentage — university 41%, office 21%, hospital 18%
"What competes": competitor density in range, including flagging a suspicious absence of it
The real places within walking distance, with metres — the score can be checked, not just believed
Ingestion from OpenStreetMap and Overture Maps, deduplicated across sources
H3 geospatial indexing and an interactive map with a per-cell inspector
Stack
Fastify
PostgreSQL
H3
DuckDB
React
MapLibre
These are our own products, not client work — we can show the architecture and the decisions behind them without breaching anyone's confidence. Client projects we discuss on request, under NDA.
Technology
What we build with
We pick mature technology with a large community and people available on the market — so you are never locked in to us.
Frontend
Next.js
React
TypeScript
Tailwind CSS
Vite
Mobile
React Native
Flutter
Expo
Swift
Kotlin
Backend
Node.js
NestJS
Fastify
Python
FastAPI
.NET
Data
PostgreSQL
Redis
ClickHouse
Elasticsearch
S3
Cloud
AWS
Microsoft Azure
DigitalOcean
Cloudflare
Infrastructure
Docker
Linux
Nginx
GitHub Actions
Grafana
AI
Claude API
OpenAI
pgvector
LangChain
Whisper
Tell us what you want to build
Three steps, a few minutes. You get a code to follow your request and the project's progress in real time.
