I'm an AI/ML Engineer with 2+ years of experience delivering intelligent systems for international clients. My work spans the full stack of modern AI engineering — from fine-tuning multimodal LLMs with LoRA/QLoRA, to building hybrid RAG pipelines with pgvector, to shipping full-stack AI products in production.

• 🔭 Currently building production RAG systems and agentic workflows for clients via Upwork
• 🧠 Specializing in LLM fine-tuning, retrieval-augmented generation, and AI agents
• 🚀 Shipped 25+ AI projects end-to-end — from architecture to deployment
• 📍 Based in Karachi, Pakistan

Languages

Gen AI & LLMs

Fine-tuning

Vision & CV

Cloud & Stack

Databases & Search

Full-stack finance app for the Pakistani market built with Next.js + Supabase + Groq

• Parses bank SMS from HBL, UBL, MCB, and Easypaisa automatically
• Urdu voice input, receipt OCR, RAG-powered financial chat
• 90%+ AI categorization accuracy at zero monthly infrastructure cost
• Deployed on Vercel + Supabase free tier

Next.js TypeScript Supabase Groq Hugging Face Vercel

End-to-end fine-tuning across 4 model architectures on custom vision-language datasets

• Fine-tuned Gemma, LLaVA, iDefics, and InternVL using LoRA/QLoRA
• Architecture-specific adapter configs; covers dataset prep → eval → inference
• Parameter-efficient methods selected based on memory constraints and architecture

Hugging Face PEFT LoRA/QLoRA Gemma LLaVA iDefics InternVL Unsloth

Production-grade retrieval-augmented Q&A over medical PDFs and clinical documents

• Hybrid search (dense + sparse) with semantic chunking and reranking
• Citation-grounded generation to minimize hallucination on clinical queries
• pgvector backend with FastAPI serving layer

LangChain LlamaIndex pgvector OpenAI FastAPI

Agentic pipeline for autonomous LinkedIn post generation

• Multi-step reasoning: web research → synthesis → tone control → hashtag optimization
• Tool-use loop with structured output validation
• Built with LangChain Agents and Serper API

LangChain Agents OpenAI Serper API Python

Automated document field extraction using Google Document AI

• Entity extraction and structured data output from unstructured documents
• Part of a broader RAG + extraction workflow reducing manual processing by ~70% Google Document AI Python OCR n8n

• 🎓 BS Software Engineering — University of Karachi, UBIT (2020–2024)
• 📜 Machine Learning Specialization — DeepLearning.AI / Coursera
• 📜 Google Data Analytics Certificate — Google / Coursera