📍 Karachi, Pakistan | 🎓 Computer Systems Engineering @ NED University | muneeb.me
I’m an execution-focused Applied AI Engineer and Full Stack Developer. I don't just dwell on theory; I craft intelligent digital experiences and production-ready systems. From low-level C/C++ performance tuning to architecting advanced RAG pipelines, fine-tuning LLMs, and building robust B2B/B2C platforms. I focus on building real-world solutions that prioritize cost-efficiency, low latency, and clean architecture.
- AI & Agentic Systems: LLM Fine-tuning (Unsloth, LoRA), RAG Architectures, LangChain, Pinecone, Chroma, Custom MCP Servers.
- Software & Web: MERN, Next.js, React, Node.js, FastAPI, Flutter, Qt5.
- Infrastructure & DB: PostgreSQL, MongoDB, Vector Databases, AWS, GCP, Docker, Linux (Fedora/NixOS/Arch).
(Click to expand and explore)
🤖 Applied AI & Agentic Solutions
- Funnel2Tunnel: Maritime Domain Expert Chatbot. A SaaS trained on 800+ maritime documents with OCR scanning and multichat capabilities. Built with Next.js, LangChain, Pinecone, and AWS.
- Chat with SQL: Locally deployable Text-to-SQL pipeline enabling natural language interaction with PostgreSQL using Gemma 3 1B and Chroma.
- iura (Mera Waqeel): A bilingual agentic AI legal assistant for Pakistan utilizing RAG and a fine-tuned LLaMA model.
- SmileReveal: AI-powered dental simulation and shade-matching mobile web app.
- Swedish Voice Agents: Real-time AI call automation integrating Twilio and OpenAI.
🌐 B2B SaaS & Cloud Infrastructure
- TIE Tools (CRM BI Platform): Built for organizations to automate reporting by integrating data from Google Ads, Meta Ads, and GoHighLevel. Features specialized tools for QA, Media Buyers, and Account Managers.
- Eggsplain Cloud: A complete cloud infrastructure management SaaS platform built with React.
- Fluuz Bank: Next-gen crypto payment and dashboard solution for Web3 fintech.
👁️ Computer Vision & Deep Systems
- Brain-Controlled Wheelchair: An end-to-end system mapping EEG signals through an ML pipeline to an ESP32 for real-world wheelchair control.
- Brain Tumor Segmentation: Utilized UNET architecture achieving ~80% IoU.
- Vehicle Detection & Re-ID: Deep pipeline built with YOLOv8, DeepSORT-style tracking, and ResNet.
- Systems Engineer @ Eggsplain (Dec 2025 - Present): Developing custom SaaS and white-label marketplace apps for hardware automation. Built the core architecture for TIE Tools to aggregate intelligent data and save employee time.
- AI Developer @ TekRevol (2025): Engineered localized LLM fine-tuning pipelines. Leveraged Unsloth to achieve 4x GPU efficiency and 100x faster LoRA training, improving overall training efficiency by 40%.
- Web & AI Solutions Architect @ Freelance / MakeShip.dev (Mar 2022 - Present): Delivering cost-optimized agentic solutions, autonomous bots, and SaaS platforms for global clients on Fiverr and Upwork.
When I’m not jumping between low-level systems and AI pipelines, I’m usually dialing in my jump rope routine, diving into the A Song of Ice and Fire universe, or exploring the streets of Disco Elysium.
Let's build something scalable.
