Book a call with the founders to talk about Rthmn
The 100ms
Budget
Engineering for human perception. Why 100 milliseconds is the boundary between "instant" and "waiting", and how we stay under it.
Human Perception
The human brain perceives delays differently at different thresholds. Understanding these boundaries defines our performance targets.
User feels in control. Cause and effect are directly connected. This is our target.
User notices delay but stays engaged. Acceptable for non-critical interactions.
User loses focus. Mind wanders. In trading, opportunities are missed.
Spending the Budget
100ms isn't much. Every component gets a slice, and there's no borrowing from tomorrow.
Price update arrives
↓ 15ms network
Process & calculate
↓ 5ms processing
Internal service hop
↓ 20ms network
Broadcast to client
↓ 15ms network
Render update
↓ 8ms render
Total: ~73ms ✓ Under budget
How We Got There
Every millisecond was earned through specific optimizations.
JSON → Protobuf
Binary serialization eliminates parsing overhead. 10x improvement in serialization time.
WebSocket → gRPC
HTTP/2 multiplexing and persistent streams reduce connection overhead.
Node → Bun
Native TypeScript execution and optimized V8 bindings.
React Renders
Memoization, virtualization, and surgical state updates.
Staying Under Budget
Performance isn't a one-time achievement. It's a constant discipline.
Performance Budgets
CI/CD fails if P95 latency exceeds threshold. No exceptions, no overrides.
Real User Monitoring
Synthetic benchmarks lie. We measure from real users, real networks, real devices.
Instant Alerts
Latency spike at 3am? We know about it before our users wake up.
Every Millisecond Matters
In real-time trading, latency isn't just a technical metric, it's the difference between seeing an opportunity and missing it. We obsess over milliseconds so our users don't have to.