Skip to content
View salahuddinuqaili's full-sized avatar
5 followers · 15 following

Achievements

Achievement: Pair Extraordinairex4Achievement: YOLOAchievement: Pull Sharkx3Achievement: Quickdraw

Achievements

Achievement: Pair Extraordinairex4Achievement: YOLOAchievement: Pull Sharkx3Achievement: Quickdraw

Block or report salahuddinuqaili

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
salahuddinuqaili/README.md

Salahuddin Uqaili

Associate Product Manager at Delivery Hero in Berlin, working on logistics CX products at scale. Outside of work, I explore what a PM can build when AI writes the code — I design the architecture, make the trade-off calls, and direct AI agents to handle the implementation. Each project is an exploration that's taught me how AI orchestration, MCP servers, and agentic workflows actually work under the hood.

What I've explored

Pulse — GPU performance monitor for systems running gaming and local AI workloads. Still in design/spec phase.

  • Chose Tauri 2 over Electron after building Neon — documented the tradeoffs
  • Rust backend for direct hardware access via NVML
  • Real-time data patterns: ring buffers, tiered polling by data volatility

Neon Protocol IDE — Desktop IDE for navigating codebases through visual architecture maps and conversational AI. 20 releases shipped.

  • IPC as the security boundary — API keys never touch the renderer process
  • Multi-provider LLM routing with priority fallback for zero vendor lock-in
  • Zustand over Redux — 6 domain slices, independently testable
  • Most complete exploration: 9,800+ LOC, 38 IPC handlers, 8 documented ADRs

Skillich — 1,028 skills across 88 roles, each rated for AI impact. Python SDK with MCP server, OpenAI/Anthropic adapters, and CLI. Published on PyPI.

  • Agent-first architecture — AI tools are the primary consumers, not humans
  • MCP server design with multi-format function calling (OpenAI/Anthropic/MCP adapters)
  • YAML taxonomy over a database so contributors can submit PRs, not SQL

RAG Starter — Your first RAG pipeline. Local with Ollama, no frameworks, no API keys, 15 minutes. 42 tests.

  • Framework-free by design — no LangChain or LlamaIndex, so every line is readable
  • Hand-written recursive text splitter (most RAG failures stem from chunking)
  • Groq free tier for Colab — beginners shouldn't need a credit card

What ties them together

  • AI-directed development — every project built by orchestrating Claude Code, not manual coding
  • Documented trade-offs — each repo has a DECISIONS.md explaining what was chosen, rejected, and why
  • Local-first architecture — all projects work without cloud dependencies
  • Multi-interface design — CLI, web, and agents share a single core (no logic duplication)

Connect

LinkedIn

Pinned Loading

  1. neon-protocol-ide neon-protocol-ide Public

    Desktop IDE for navigating codebases through visual architecture maps and AI. Built entirely through AI-directed development. 20 releases shipped.

    TypeScript 4 3

  2. pulse pulse Public

    GPU performance monitor for gaming + local AI workloads. See what your GPU is doing, know what it can still do. Tauri 2 + Rust.

    TypeScript

  3. rag-starter rag-starter Public

    Your first RAG pipeline. On your own documents. 15 minutes. Try it now in Google Colab — no install needed.

    Python

  4. skillich skillich Public

    Know exactly which skills to invest in to become irreplaceable in the age of AI. 88 roles, 1,028 skills, each rated for AI impact.

    Python