Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a CV model for image segmentation using pytorch
Data Scientist and ML Engineer building production AI and deep learning systems
I build production ML/DL systems — not prototypes or demos.
Research Assistant at MBZUAI (UAE): AI cancer cell detection using CellViT++ deep learning on histopathology data.
Delivered: RAG pipeline w...
About this Gig
Most computer vision freelancers give you a notebook that works on their machine. I give you a production pipeline that works on yours.
I am a Research Assistant at Punjab University Lahore, where I built a cancer cell detection system on histopathology slides using CellViT++ and deep learning. Manuscript in preparation for journal submission.
WHAT I CAN BUILD FOR YOU
Image classification and object detection
Image segmentation (U-Net, SegFormer)
Medical image analysis (histopathology, X-ray, MRI, CT)
Cell detection and counting pipelines
Grad-CAM explainability visualizations
FastAPI deployment + Docker containerization
TECH STACK
PyTorch · OpenCV · U-Net · SegFormer
CellViT++ · YOLO · FastAPI · Docker
Streamlit · albumentations · MONAI
HOW TO START
Message me with:
1. Your images or dataset
2. What you want detected or segmented
3. Your deadline
I will recommend the right package before you order.
Medical imaging and research collaborations welcome.
APIs:
Other
Programming language:
Python
•
R
•
MATLAB
•
SQL
Tools:
Jupyter Notebook
•
OpenCV
•
TensorFlow
•
MLflow
•
PyTorch
Frameworks:
Scikit-learn
•
SimpleCV
•
Keras
•
PyTorch
•
Panda
My Portfolio
Other Data Science & ML Services I Offer
FAQ
Why not just use a free model from Hugging Face?
Pretrained models need fine-tuning on your data, evaluation on your classes, and a deployable API. I handle the full pipeline — training, metrics, and FastAPI endpoint. That's what you're paying for, not the base model.
0 Do you have experience with medical or scientific images?
Yes. I worked as a Research Assistant, building a cancer cell detection model on histopathology slides using CellViT++. Manuscript in preparation. I apply the same research-grade methodology to every project.
What exactly do I receive on delivery?
Model weights, full source code, performance report (IoU/Dice/AUC), README, and Grad-CAM visuals. Standard and Premium include FastAPI endpoint. Enterprise adds Docker and cloud deployment. No black boxes.
I don't have a labeled dataset — can you still help?
Yes. I can source a suitable public dataset and adapt it via transfer learning. If your data is unique, I'll advise on labeling strategy and build around what you provide. Message me first to confirm the approach.
What if the model doesn't perform well enough?
I define the target metric in writing before we start. Revisions are included in every package for tuning and adjustments. If your data can't support the goal, I'll tell you during scoping — not after delivery.
