The Blend.
Custom point-of-sale and management software for a coffee shop. Built around order handling and café operations, with an offline-first order queue and synchronization described in the project documentation.
INDEPENDENT THINKING. APPLIED ENGINEERING.
PAKISTAN · PORTFOLIO / 2026
AI Engineer & Software Developer
Muhammed
Ahsan.
From a promising experiment
to something people can use.
I build computer vision applications, research datasets and end-to-end software. Curious about the model. Committed to the whole system.
ONLINE AS @ahsanneural ↗
3D companion unavailable on this device. The portfolio is ready to explore.
01 / ABOUT
Intelligence is useful
when it works in the world.
I’m Muhammed Ahsan, an AI engineer and software developer studying Artificial Intelligence at The Islamia University of Bahawalpur.
My work connects machine learning experiments with practical software: inspection tools, AI web applications and systems built for a client’s everyday operations.
I’m particularly interested in computer vision and the data behind it. Alongside development, I curate public datasets and contribute to research on model evaluation.
More about my journey ↗Kaggle Datasets Grandmaster
View
profile ↗
Public dataset curation across applied domains.
02 / SELECTED WORK
Ideas, implemented.
A selection of public engineering work.
Built across AI and software.
Showing 4 projects
Steel Defect Vision
A YOLOv8s inspection application combining NEU-DET and GC10-DET data. Supports single-image and batch inspection, adjustable detection thresholds and annotated outputs.
PCB Defect Detection
A browser-based inspection tool for six PCB manufacturing defect types, using YOLOv8s and DeepPCB. Includes visual summaries, configurable thresholds and CSV reports.
Ivonyx AI
An AI web application with chat and content-generation workflows, persistent conversations and generation history. Developed during an AI engineering internship; proprietary application code and deployment remain private.
03 / EXPERIENCE
Learning through
the work itself.
JUN — SEP 2026
REMOTE
REMOTE
Hexovate Solutions
AI / ML Intern
Worked with cross-functional teammates on practical AI/ML projects and model-building workflows. Applied academic machine learning knowledge to implementation, experimentation and deployment-oriented work.
Public work is featured above. Internal and client-specific employer details remain confidential.
INDEPENDENT
CLIENT DELIVERY
CLIENT DELIVERY
The Blend.
Full-Stack Developer
End-to-end solo development of the coffee shop’s POS application, covering the application workflow, interface and live delivery.
04 / RESEARCH & PUBLICATIONS
ASK BETTER QUESTIONS. BUILD BETTER MODELS.
Beyond the
headline accuracy.
Subgroup-Disaggregated Fairness Audit of a Dermatoscopic Lesion Classifier
S. A. Rehman, M. Ahsan & M. U. Khan
Research on subgroup-disaggregated evaluation in medical computer vision.
Read the preprint ↗ 10.2139/ssrn.748461805 / DATASETS
Good AI starts
with good data.
Curated datasets for exploration,
experimentation and research.
01
IMAGE DATA
Rare Neurological Diseases MRI
Curated Edition · medical imaging
02
STRUCTURED DATA
Rare Diseases — Orphadata 2026
Structured rare-disease data
My public collection also covers food and nutrition, jobs and skills, traffic accidents and climate.
Explore datasets on Kaggle ↗06 / TECHNICAL EXPERTISE
A model is only
part of the system.
- 01 / Applied AI & vision
- Python · YOLOv8 · OpenCV · scikit-learn · Model evaluation · Data preprocessing
- 02 / Product engineering
- TypeScript · JavaScript · Next.js · React · REST APIs · Streamlit
- 03 / Data & delivery
- SQL · PostgreSQL · Prisma · Supabase · Pandas · NumPy · Git · Vercel · Cloudflare Pages
07 / EDUCATION & TRAINING
Always a student.
IN PROGRESS
BS Artificial Intelligence
The Islamia University of Bahawalpur
Additional training
ML Summer School 2026Cohere LabsAUG
2026
Computer Vision FoundationsUltralyticsAUG
2026
08 / CONTACTBACK TO TOP ↑
Let’s make
something useful.
Engineering opportunities, thoughtful collaborations
and interesting problems.