Skip to content
aiaz·labs

Labs · research notes

Findings from the lab.

Experiments, measurements and findings from building AI interfaces, RAG pipelines, accessibility and performance work. Each note records what was tried, what was measured and what held up.

Featured
AI Interfaces4 min read

Building a streaming chat UI that doesn't jank

Token-by-token rendering, scroll behaviour and abort handling — the patterns that keep an LLM chat feeling instant.

  • Streaming
  • React 19
  • UX
Read the note

All posts

Filter by the topic you care about.

Showing 6 posts

  • AI Interfaces4 min read

    Building a streaming chat UI that doesn't jank

    Token-by-token rendering, scroll behaviour and abort handling — the patterns that keep an LLM chat feeling instant.

    • Streaming
    • React 19
    • UX
    Read the note
  • RAG6 min readComing soon

    Citations users can actually verify

    How to ground RAG answers and link every claim back to the exact source passage.

    • RAG
    • Grounding
    • Citations
    In the works
  • Accessibility7 min readComing soon

    Keyboard-first comboboxes: ARIA 1.2 in practice

    What the spec says, what screen readers really announce, and the bugs that slip through.

    • ARIA
    • Keyboard
    • WCAG 2.2
    In the works
  • Optimised9 min readComing soon

    Rendering 100k rows at 60fps

    Windowing, memoisation and measuring with the profiler — a practical guide to fast data grids.

    • Performance
    • Virtualisation
    In the works
  • RAG10 min readComing soon

    Hybrid search: when keywords beat embeddings

    Combining BM25 with vector search and reranking, and how to tell which one is winning.

    • Hybrid search
    • Rerank
    • Evaluation
    In the works
  • Accessibility5 min readComing soon

    Announcing changes: aria-live without the noise

    Polite vs assertive, toasts, async results — making dynamic UI usable with a screen reader.

    • aria-live
    • Screen readers
    In the works