AI Engineer &
Full Stack Developer.
Hi, I'm Muhammad Abdullah. I build LLM systems, RAG, Document AI, and production full-stack platforms with Python, modern .NET, and cloud engineering.
Selected Works
AI systems and full-stack products, from RAG and Document AI to enterprise platforms.
Agentic AI Portfolio Assistant
LangGraph/Next.js agent with Groq GPT-OSS-120B, Weaviate hybrid/BM25 RAG, structured tool calling and a deterministic lead-management state machine.
AutoML / Neural Architecture Search
Evolved neural architectures for user datasets using genetic selection, crossover and mutation over backbones, layers, neurons and activations.
ALMS HRM System
Full-featured HRM for FAST CFD Campus: attendance, payroll, employee data, biometric iFace integration.
Brail App
Mobile app to interpret Braille via camera + CNN model; optimized for mobile with TensorFlow Lite.
Face Recognition Attendance System
Automated attendance from classroom group photos using adaptive histogram equalization, YOLO face detection, GAN super-resolution and FaceNet embeddings.
Virtual Trainer
AI fitness assistant analyzing posture and counting reps via skeleton-based keypoints.
Latest Insights
Thoughts on software engineering, AI agents, and building scalable systems.
Experience
AI systems, full-stack delivery, and enterprise engineering.
Feb 2021 - Present
Feb 2021 - Present
Senior Software Engineer
Glosix Systems
AI-Assisted Application Modernization Pipeline
- Architected and built an AI-assisted application modernization pipeline that enabled completion of a .NET Framework 4.8 to .NET 10 migration in approximately 1 month, versus an estimated 6-8 months of manual engineering effort.
- Automated structural code transformations in Python; built module scanners and generated dependency-aware prompts with file paths, function checklists and migration plans, focusing LLM reasoning on business logic and AppService implementations.
Document AI / Intelligent Document Processing
- Built a Document AI pipeline achieving 96%+ structured field-extraction accuracy for UK Penalty Charge Notices (PCNs), using image preprocessing, PaddleOCR, optional GLM-OCR, Groq/Ollama and deterministic validation.
- Engineered structured JSON extraction, asynchronous jobs, multi-tenant storage, restart recovery and human review; combined PDF conversion, CLAHE enhancement and deskewing with field normalization and OCR retries.
RAG Knowledge Assistant
- Built a retrieval-augmented generation (RAG) assistant with FastAPI, Ollama/Llama 3, SentenceTransformers and Weaviate; implemented PDF/DOCX/text/web ingestion, chunking, generated Q&A indexing, HNSW semantic retrieval and conversation history.
Enterprise Client Delivery | SpeedAutoSystems
- Led production implementations for 10 large clients and many small to medium clients, including Europcar/Gargash (UAE), Al Jazirah and Al Jabr (Saudi Arabia); worked directly with clients from requirements and architecture through implementation, rollout, production support and ongoing delivery.
- Built and delivered end-to-end integration pipelines for Tajeer (Saudi Arabia) and TARS/RTA (Dubai).
- Owned legacy client migrations into SpeedAutoSystems, from schema analysis and mappings to transformations, custom .NET ETL, dependency handling and bulk data migration.
- Managed Azure infrastructure and production deployments across App Services, Functions and Elastic Pools; mentored developers and guided architecture and implementation decisions.
Feb 2018 - Mar 2021
Feb 2018 - Mar 2021
Lead Developer
FAST-NUCES University
University HRM and Gate Pass
- Led architecture and delivery of an ASP.NET MVC university HRM system with iFace 800/1000 attendance integration.
- Built a Flutter Android gate-pass app; coordinated delivery, mentored developers and guided software design and code quality.
Feb 2017 - Jan 2021
Feb 2017 - Jan 2021
Software Engineer
Freelancing
Web, Mobile, Desktop and AI Delivery
- Delivered 50+ projects across web, mobile, desktop and AI using Python, React, JavaScript, ASP.NET MVC, Flask, Django and TensorFlow.
- Built a Basketball Match Analyzer combining player/ball detection, multi-object tracking, facial identification, pose estimation and movement-based activity classification.
- Engineered an Exercise Form Checker & Rep Counter using body keypoints, geometric joint angles, reference pose comparisons and movement-state transitions to validate form and count repetitions.
Technical Expertise
AI engineering and full-stack delivery — LLMs, computer vision, .NET, Python, and cloud infrastructure.
Production AI
and full-stack systems.
I’m an AI Engineer and Full Stack Developer with 7+ years spanning LLM systems, RAG, Document AI, computer vision, and enterprise ERP. I shipped an AI-assisted .NET migration in about a month instead of 6–8, and a PCN Document AI pipeline at 96%+ extraction accuracy.
Day to day I combine Python, modern .NET, and cloud engineering — leading SpeedAutoSystems production delivery for UAE and Saudi Arabia rental clients, plus FastAPI RAG assistants, Azure, and ETL at scale.
Education
Academic background & foundations.
FAST NUCES
Bachelor of Computer Science
Aug 2016 — Aug 2020
Punjab College
FSc Pre Engineering
Mar 2014 — Mar 2016
Let's Work Together
I'm always open to discussing new projects, creative ideas, or opportunities to be part of your vision.