In this talk, I’ll share what I learned while building tanstack-faster, a high-performance e-commerce template powered by TanStack Start, TanStack Router, TanStack Query, and the Cloudflare edge ecosystem, inspired by NextFaster.
I’ll walk through the architectural decisions, caching strategies, image optimization techniques, and intelligent preloading patterns that help real-world applications feel fast without driving up infrastructure costs.
This session is for engineers who genuinely care about the performance of their full-stack applications.
Problem Space
Next.js is currently the most widely adopted full-stack React framework. While it works well for many use cases, it is increasingly becoming bloated and over-generalized, especially when performance, cost, and control matter.
In this talk, I want to explore TanStack Start as a modern alternative that does not compromise on either speed or developer experience and Control, and show how developers can:
deliver instant navigations and content-rich pages,
manage complex data flows and caching across client and edge servers,
optimize heavy media like images,
keep global infrastructure costs predictable and low.
Most starter templates stop at scaffolding. tanstack-faster focuses on integrated performance patterns using type-safe full-stack React, advanced data strategies, and edge-first economics.
What You’ll Demonstrate
Real Architecture Overview
A walkthrough of the complete stack:
TanStack Start for full-stack SSR and streaming
TanStack Router for type-safe routing and intelligent preloading
TanStack Query for client-side caching and state orchestration
Cloudflare Workers, KV, D1, and R2 for edge compute, caching, databases, and asset storage
Advanced Performance Techniques
Concrete techniques used in the project:
Image Optimization
Preloading
Multi-Level Caching
Cloudflare edge cache → Cloudflare KV → browser cache via TanStack Query
Designing cache layers to balance performance, correctness, and cost
Cloudflare Edge Stack in Action
How Cloudflare enables the architecture:
Workers for global SSR and routing
D1 as a low-latency, SQLite-compatible edge database
R2 for cost-effective object storage
KV for hot data caching at the edge