I will build smart ai models for object detection, segmentation and classification
About this Gig
Do you have image or video data? Let's turn it into an AI model that actually does something useful.
I'm a Mechatronics Engineer with real research experience in AI and robotics, not just tutorials and YouTube courses. I've trained, debugged, and deployed models for classification, detection, tracking, and more.
What I build:
- CNN image classifiers with Grad-CAM heatmaps so you can see exactly what the model is learning
- YOLO-based object detection and tracking, fine-tuned on your data
- End-to-end pipelines from raw data to cloud deployment with API integration
- Trained model file and usage guide included in every package.
- Full source code in Premium.
I won't just hand you files and disappear. I'll make sure you understand what you got and how to use it. Clear communication, no ghosting, and if something isn't right we fix it.
Not sure which package fits? Just message me first.
I'd love to hear about your project!
Programming language:
Python
•
MATLAB
Tools:
Jupyter Notebook
•
OpenCV
•
TensorFlow
•
CVAT
•
Colab
•
PyTorch
Frameworks:
Scikit-learn
•
Google ML Kit
•
Keras
•
PyTorch
•
Panda
FAQ
What languages do you speak?
I speak English professionally and Urdu natively, so communication is never an issue regardless of where you're from.
What is your educational background?
I hold a 4-year Bachelor's degree in Mechatronics Engineering, which covers robotics, AI, electronics, and software all in one.
How much experience do you have in AI and Robotics?
About 3 years across academic projects, personal builds, and hands-on research in AI and robotics.
How many projects have you worked on?
Between 10 and 20 projects so far, ranging from personal experiments to academic research and real-world applications.
How many hours a week are you available?
I put in 40+ hours a week, so you can expect fast responses and consistent progress on your project.
Can I verify your profile somewhere outside Fiverr?
Sure! Share your concerns in our chat and I'll send you my LinkedIn and GitHub directly. If you're still not satisfied and want to leave, that's completely fine. I'm here to satisfy my clients, not pressure them.
Do I need to provide my own dataset?
Ideally yes, but if you don't have one I can help source or build a suitable dataset as part of the project.
What format will the trained model be delivered in?
Usually .pt or .pth depending on the framework, along with a clear guide on how to load and run it.
Can you work with a custom or niche dataset that isn't publicly available?
Absolutely, that's actually where custom training shines most. Just share the details and we'll figure it out.

