AI Engineer | AI Agent Engineer
Building intelligent systems across the full AI spectrum — from Machine Learning and Deep Learning to LLMs, RAG pipelines, and Agentic AI automation.
I'm Adeel Iqbal, an AI Engineer with experience spanning the full AI spectrum — from training classical Machine Learning and Deep Learning models to building production LLM applications, RAG pipelines, and Agentic AI systems.
I work across Computer Vision, NLP, Generative AI, and multi-agent architectures. Currently, I design AI automation and intelligent agent systems for real business clients, helping them replace manual workflows with fully autonomous pipelines.
For me, AI isn't just about models or benchmarks — it's about building things that actually work in the real world and create tangible impact.
LangGraph multi-agent system that generates tailored cover letters, categorized interview Q&As, and CV gap analysis with severity ratings. Features a full HITL workflow with MemorySaver state persistence and a Chainlit streaming interface.
Enterprise LangGraph multi-agent platform for Saylani Mass IT Training — handles natural language database queries, automated bulk email and WhatsApp communications, and real-time student management. Processes 100+ parallel emails in 15 seconds.
24/7 AI voice receptionist that handles appointment booking, rescheduling, and cancellations with zero human intervention — saving 15-20 hours of staff time per week. Powered by VAPI and ElevenLabs with n8n automation for calendar and email workflows.
AI WhatsApp ordering agent that guides customers through the menu like a human waiter, collects cart-based orders with variations and add-ons, verifies payments, and tracks deliveries — built with n8n, OpenAI, and Airtable.
Multi-agent RAG system that ingests research PDFs and generates platform-native content for LinkedIn, Instagram, YouTube, and X. Uses hybrid search with dense vectors, BM25, and cross-encoder re-ranking across 7 domain collections, with a React 19 frontend.
End-to-end n8n automation monitoring the HR inbox 24/7. Uses GPT-4o to score candidates 0-100 against job descriptions, cutting screening from 10-15 minutes to under 30 seconds per resume. Auto-logs to Google Sheets and alerts HR for top candidates.
Production-ready RAG chatbot that crawls any website, builds a vector database from the content, and delivers AI-powered customer support with precise, context-aware responses grounded in the site's actual data.
n8n workflow that scrapes competitor social media, analyzes post performance, and generates AI-powered branded content with multilingual support and platform-specific recommendations.
Real-time drone detection using YOLOv11n with 94.3% mAP@50 accuracy. Highlights drones with bounding boxes in images and videos via a FastAPI web interface.
Deep Learning system for automated Alzheimer's detection from Brain MRI scans, achieving 96.9% accuracy using fine-tuned ResNet50 on the OASIS dataset.
Combines YOLOv11 for detection and SAM2 for precise segmentation of brain tumors in MRI scans, delivering accurate boundary mapping for medical imaging workflows.
CNN model with 98.9% accuracy that detects 4 major rice diseases from leaf images, helping farmers identify crop health issues early for timely intervention.
ANN model for flood probability prediction with preprocessing, feature importance analysis, and a real-time risk assessment input function for disaster preparedness.
ML app that predicts high or low pregnancy risk using a tuned Random Forest model, designed for early detection and improved maternal health outcomes.
Recommendation system using Content-Based and Popularity-Aware Filtering, built with Streamlit for an interactive music discovery experience similar to Spotify's own engine.
Deep Convolutional GAN trained to generate realistic anime character faces from noise, implementing the full GAN training loop with discriminator and generator networks from scratch.
Sequence-to-sequence Transformer built from scratch for language translation, implementing multi-head attention, positional encoding, and the full encoder-decoder architecture.
Android keyboard that converts Roman Urdu, Hinglish, and broken English into fluent natural English in real time using GPT-4o Mini — with suggestions firing 1.5 seconds after you stop typing and a one-tap revision flow built for Pakistani and South Asian users.
Android app that turns any Upwork job post into 3 ready-to-send, high-converting proposals in seconds using GPT-4o — built for freelancers who want to win more clients with less time spent writing bids.
Want a full overview of my experience, skills, and projects? Download my CV below.
Download CVOpen to new opportunities, freelance projects, and interesting collaborations. Feel free to reach out.
Karachi, Pakistan
adeelmemon096@yahoo.com
linkedin.com/in/adeeliqbalmemon