Systems thinking
Realtime Interface Engineering
Root-caused a 1GB memory leak on a live trading platform — incorrect useEffect dependency causing cascading re-renders on every WebSocket tick.
Hey, I'm Trish
Building frontend systems. Debugging where they break.
I build and debug performance-critical, real-time frontend systems. My work lives where state, rendering, and correctness collide — trading platforms, healthcare products, and long-running UIs that fail in non-obvious ways.
I write about frontend engineering the way it actually behaves in production — not the way tutorials describe it.
Technologies I work with
ReactReact NativeNext.jsTypeScriptNode.jsWebSocketsReduxTailwind CSSDocker
60%
Memory Reduction
1GB → 400MB on a live trading platform
82%
Bundle Size Cut
Sentry SDK from 400KB to 70KB gzipped
500K+
Users Migrated
Zero-downtime geo-based platform migration
100K+
Traders Served
Position analytics with real-time market data
Approach
What I optimize for
Failure analysis
Production Debugging
Restructured Sentry SDK initialization and tree-shook wildcard imports to cut monitoring bundle from 400KB to 70KB — an 82% reduction.
Scale
Performance Architecture
Led geo-based routing and compliance gating for 500K+ existing users and 600K+ new sign-ups across jurisdictions — zero downtime migration.
Writing
Latest essays
March 11, 2026
Why Sentry Was Showing 11-Second Button Clicks: Debugging Bad Data, Multiple Bundles, and an 82% Size Cut
Garbage Sentry data led us to multiple SDK bundles and wildcard imports. Fixing the init pattern and tree-shaking cut our Sentry payload from 400KB to 70KB.
Read essay
February 18, 2026
How I Cut a Trading Platform's Memory from 1GB to 400MB
A production debugging story — how I tracked down a 1GB memory leak on a crypto trading platform caused by a subtle useEffect dependency and unmemoized table re
Read essay
Deep work
What I work on
The kinds of production problems I spend most of my time solving — real incidents, not hypotheticals.
01
Memory leaks under live WebSocket traffic
Traced a 1GB heap bloat on a trading page to an incorrect useEffect dependency causing cascading re-renders on every market tick. Reduced to 400MB.
02
Monitoring bundles that silently break observability
Sentry SDK was loading 400KB gzipped due to wildcard imports and late initialization. Restructured to tree-shakeable named imports — 70KB, error tracking from first paint.
03
State accumulation in long-running chart surfaces
Options analytics page crashing for 15% of users. Root cause: unbounded state growth in chart re-renders under sustained market data updates.
04
Cross-platform consistency at scale
Built a design system using Amazon Style Dictionary with custom transforms, adopted 100% across React Native and web teams, cutting design-to-code drift by 60%.
Stack
Skills & tools
I optimize for depth in a few areas rather than shallow familiarity across many tools.
Frontend
ReactNext.jsTypeScriptReduxZustandTailwind CSSFramer MotionRadix UI
Mobile
React NativeExpoReanimatedFCMEAS Build
Backend & Infra
Node.jsExpressPostgreSQLRedisWebSocketsSocket.IODockerGCP
Tooling & Monitoring
ViteWebpackSentryApache EChartsRecharts
Philosophy
How I think
Frontend systems are long-running, stateful machines — not just collections of components.
Most serious UI failures I've debugged were not visual issues. They emerged from performance regressions, unstable references, or incorrect assumptions about state and timing.
These problems often surface as broken UX, even though the root cause lives deep in rendering behavior, data flow, or lifecycle management.
Now
1Shipping a mental health platform to production users
2Writing deep technical essays on frontend failure modes
3Building cross-platform React Native experiences
4Exploring WebSocket reliability under unstable networks