I'm an AI/ML Engineer and AI Researcher based in London, building agentic AI systems, computer vision pipelines, and LLM-powered automation for healthcare, finance, and enterprise workflows. I hold an MSc in Artificial Intelligence Technology (Distinction) from Northumbria University, London, and I work across the full ML lifecycle — from data pipelines and model development to containerised deployment and cloud infrastructure.
- 🔬 Currently researching multi-modal agentic systems for medical AI and graph learning for financial crime detection
- 🌱 Deepening expertise in Vision Transformers, Multi-Agent Orchestration, and Weak/Positive-Unlabelled Learning
- 🤝 Delivered end-to-end AI solutions for international clients across healthcare, automotive due-diligence, and banking compliance
- 📫 Reach me at malikharis3984@gmail.com
| Publication | Venue | Status |
|---|---|---|
|
Agentic AI for Multi-Modal Cancer Detection (MSc Thesis) Agentic AI architecture combining computer vision models with LLM-orchestrated reasoning pipelines for multi-modal cancer detection, with anomaly-detection-driven data validation across multiple medical imaging datasets. |
Wiley — Health Science Reports | Under Review |
|
Intermediary-Aware Detection of Obfuscated Dual-Use Goods Procurement Networks in Banking via Expert-Informed Simulation, Weak Supervision, and Positive-Unlabelled Graph Learning In collaboration with Standard Chartered Bank |
IEEE / ACM CMT | Under Review |
Machine Learning & AI
Software Engineering
Automation & Workflow
Cloud, Data & MLOps
Visualisation & Analysis
| Role | Company | Duration |
|---|---|---|
| AI Developer Intern | Nexcell Solution, London | Nov 2025 – Mar 2026 |
| Computer Vision & AI Engineer | Neural Sight AI, New York (Remote) | Mar 2024 – Apr 2025 |
| Freelance AI/Software Engineer | Fiverr & Upwork, Remote | Jun 2022 – Jan 2025 |
| Machine Learning Intern | KSOFT, Islamabad, Pakistan | Jun 2023 – Aug 2023 |
📌 Highlights: reduced deployment time by 50% via containerised CV pipelines · improved inference performance 30–40% through systematic profiling · built sentiment models at 85%+ accuracy · delivered production ML/automation systems end-to-end for international clients.
| Project | Description | Stack |
|---|---|---|
| AML Transaction Monitoring & SAR Drafting Assistant | Agentic pipeline for detecting suspicious transaction networks and auto-drafting Suspicious Activity Reports. | Python · LLMs · Graph Learning |
| PlateCheck – Vehicle Due-Diligence Automation | n8n + React automation pulling DVSA MOT records, scoring fraud/mileage risk, and generating AI research-backed reports via web/WhatsApp. | n8n · OAuth2 · React · Twilio |
| Cancer Detection Agentic System | MSc thesis project — agentic AI combining CV models with LLM-orchestrated reasoning for multi-modal cancer detection. | PyTorch · GCP · Agentic AI |
| MedFlowAI – Healthcare Management System | Full-stack platform with OpenAI-powered diagnostics (92%+ accuracy), HIPAA-aligned Docker deployment, RBAC. | Flask · PostgreSQL · Docker · OpenAI |
| Voice Assistant AI | Real-time voice assistant integrating live data APIs via Groq for low-latency responses. | Groq API · Python |
| License Plate Recognition | YOLOv8 detection pipeline with structured output logging for automated accuracy benchmarking. | YOLOv8 · OpenCV |
MSc Artificial Intelligence Technology — Northumbria University, London (2025–2026, Distinction) Modules: Machine Learning on Cloud · Information Governance & Cyber Security · Programming for AI · AI for IoT · Leadership in a Digital Age
BSc Electrical Engineering — Comsats University, Pakistan (2020–2024, 2:1)
-0056D2?style=flat-square&logo=coursera&logoColor=white){kind=icon}
-0056D2?style=flat-square&logo=coursera&logoColor=white){kind=icon}
If any of my repositories were useful to you, consider giving them a star — it helps others discover the work and keeps me motivated to keep building in the open.
This profile README, and the repositories linked above unless stated otherwise within them, are shared under the MIT License. Feel free to fork the layout of this README for your own profile.
MIT License — © Muhammad Haris
Permission is hereby granted, free of charge, to use, copy, modify, and
distribute this content, subject to inclusion of the original copyright
notice in all copies or substantial portions.
