I'm an AI/ML Engineer, published researcher, and technical educator building end-to-end systems across medical imaging, robotics, IoT, NLP, and agentic AI. I presented my MRI reconstruction research in France at MIRASOL 2026.
I'm an AI/ML Engineer, published researcher, and technical educator studying Mechatronics Engineering at the Federal University of Technology Minna. I build intelligent systems that leave the notebook and reach real users, with hands-on experience in medical imaging AI, agentic systems, IoT, 3D printing, and autonomous robotics.
My physics-informed low-field MRI reconstruction research was presented at an international conference in France in 2026. I facilitate technical workshops for children and polytechnic educators, and serve as Technical Lead at Notion@FUTMinna. Earlier, I led a 13-person team building a road-hazard detection system, shipped a multilingual AI loan advisor for rural farmers, and scaled an agentic productivity bot to 65+ active users.
A quick path through where I've studied, worked, and been recognized.
Designed and fabricated physical prototypes with 3D printing, integrated sensors and embedded systems through IoT, built functional technology systems and tools, and handled hands-on electrical installation and hardware setup. Facilitated a one-month summer IoT workshop for children with an age-appropriate curriculum covering sensors, microcontrollers, and basic automation.
Building end-to-end machine learning and deep learning pipelines.
Serve as Technical Lead for the official Notion student community at FUTMinna. Drive digital productivity and knowledge-management culture, organise technical sessions, onboard members, and build Notion templates and workflows for academic and project use.
Served as Secretariat for a 5-day national-level workshop on the implementation of the National Diploma in Artificial Intelligence, working with polytechnic lecturers and education policymakers. Facilitated practical sessions walking lecturers through microcontroller simulation using TinkerCAD, enabling educators to teach IoT and embedded systems confidently in their institutions.
Delivered a production AI booking automation system that cut processing time by 60% and onboarded 20+ riders; provided bilingual (English/Hausa) support.
Grew community engagement by 40% and resolved 95%+ of customer inquiries within SLA in a high-volume environment.
Benchmarked five reconstruction paradigms (U-Net, CascadeNet, MoDL, DUN-DD, and E2E-VarNet) on real OSI² ONE 47 mT low-field MRI data using rigorous nine-fold Leave-One-Subject-Out cross-validation. The work is motivated by improving MRI accessibility in low-resource and rural settings where conventional 1.5–3T scanners are too expensive.
CascadeNet matched performance with ~15× fewer parameters. A zero-shot sweep evaluated acceleration factors from R=2 up to R=12 without retraining. Stack: PyTorch, E2E-VarNet, CascadeNet, MoDL, DUN-DD, U-Net, k-space processing, SSIM/PSNR, ISMRMRD HDF5, and Compute Canada GPUs.
MIRASOL — MICCIA 2026, France: Accepted paper and poster, co-authored with researchers from FUTMinna, University of Ilorin, McGill University Montreal, University of British Columbia, and MAI Lab Lagos.
Authors: Tolough Nelson Aondongu, Muhammad Inuwa Muhammad, Tella Adetayo, Victor Ejike Onyedim, Joseph John, Ayatullah Hanif Showunmi, Zhang Dong, Confidence Raymond, Udunna C. Anazodo, Aondona Moses Iorumbur, Sanni Henry Ananyi.
Led a 13-member team to detect road cracks, potholes, and bumps from images for smart infrastructure use cases, evaluated on precision, recall, and F1.
An autonomous indoor robot that navigates to workstations on command, with a real-time web command center for telemetry and control.
Financial literacy AI for rural farmers, processing loan documents in 8 Nigerian languages via OCR, ASR, TTS, and an LLM risk-assessment engine.
An autonomous Telegram productivity bot scaled to 65+ active student users through progress logging and peer-competition features.
A Telegram-based booking system for Togo Mobility, automating registration and payment verification — cutting manual booking time by 60%.
Benchmarked five physics-informed reconstruction paradigms on the OSI² ONE 47 mT low-field MRI dataset. E2E-VarNet achieved SSIM of 0.8212 with a +7.50 dB PSNR gain over U-Net. CascadeNet provides near-equivalent quality with ~15× fewer parameters.
MIRASOL, MICCIA Conference, France (2026)
Notion@FUTMinna (2026–Present)
RAIN x Meta AI Developer Academy Nigeria Hackathon (2025)
United Nations Academic Impact & MCN (2025)
Halima Ibrahim Babangida Foundation (2024)
Open to AI/ML roles, research collaboration, and interesting problems. Reach out directly, or use the form.
inuwamuhammad930@gmail.com +234 916 873 0302 linkedin.com/in/muhammad-inuwa-muhammad github.com/shadowboy-tech kaggle.com/muhinuwa x.com/shadowboy_AI muhammadinuwamuhammad.tech