Abdul Moiz
Full Stack AI Engineer
Engineer focused on building scalable tech, community-driven impact, and products that matter beyond the screen.
Get in touch
TechStack
Summary
I'm Abdul Moiz, a full-stack software engineer with a product-first mindset. I focus on building scalable, impactful web applications using the MERN stack and modern tooling, balancing solid engineering habits with adaptability so solutions serve both users and business goals.
On the AI side, I work with retrieval and agent-style patterns (including LangChain, LangGraph, RAG, and open-source models where it makes sense) and use assistants such as Claude together with Cursor so I can ship a credible first version quickly, then adapt the product as the market and user feedback evolve. The goal is speed without losing judgment: iterate in public, measure what matters, and refactor once the direction is clear.
Starting as an intern in 2022, my path has run through client-facing web work at 11-Seas, an intensive stretch at Headstarter improving APIs and Next.js performance for AI tooling, and TurboDebt in fintech, where performance work and analytics helped grow traffic and tighten the user journey. At Xische I shipped accessible React systems, auth flows, internationalization, and cloud-backed releases. Today at BitByteClub I lead AI-driven MVP delivery end to end—rapid validation with stakeholders, Supabase-backed infrastructure, MCP and Auth0 integrations, OpenAI and RAG with vector-backed meal recommendations, multi-tenant enterprise admin, and fine-tuning so the product stays accurate and trustworthy as it scales.
I'm also a leading member of Dev Weekends, a developer community where I have mentored over two hundred engineers through technical guidance, one-on-one sessions, and personal development support. Many have gone on to mentor others themselves, keeping that cycle of growth alive.
Alongside this, I founded Dastagahe Karam, a dignity-first initiative that has provided silent aid to thousands of underserved families, raising over 3M PKR while keeping privacy and compassion at the center.
Outside of engineering, interests in human behavior, performance, and growth help me practice empathy in technical decisions: better outcomes for products and for people. There was a time I stood where many juniors stand today, uncertain and overwhelmed by choice. Having found my focus, I stay committed to passing it forward so fewer people feel stuck where I once stood.
I bring a grounded mindset, technical sharpness, and a people-first approach to everything I work on, from products and teams to communities.
Experience

•Shipped AI-driven MVPs end to end on Supabase (Edge Functions, queues, auth, storage) with rapid iteration and feedback; integrated MCPs, Auth0, and external connectors; delivered OpenAI + RAG + vector DB pipelines that sharpened meal recommendations via LLM optimization and fine-tuning; contributed multi-tenant enterprise admin (tenant lifecycle, isolation, secure token handling).

•Built modular, reusable, and accessible React components and a custom authentication hook (session persistence, token refresh, role-based access control), improving UI consistency and developer experience.
Currently Reading

Hyperfocus
by Chris Bailey
Articles
Sharing insights, experiences, and knowledge from my journey in technology, community building, and personal growth. Each post is crafted to provide value and spark meaningful conversations.
Two Levels of Hugging Face APIs: From Quick Start to Full Control
High-level pipelines versus tokenizer-and-model control, quantization, streaming, and how the pieces fit in real LLM systems.
From Pipeline to Streaming
Loading models, tokenizers, quantization, internals, and streaming generation—bridging Hugging Face theory to practice.
Multi-Agent Systems, UI Layers, and Tool Calling
Multi-agent conversations, Gradio and Streamlit, tunneling local servers, and tool calling so models use real data.
Prompt Caching: Making LLMs Fast and Practical
Cache keys, reuse, and cost or latency wins when agents and long chats repeat similar prefixes.




