Aman S.

Key Skills

Software

Roboflow

Top Subject Matter

Health care

Object Detection

E-commerce

Top Data Types

Image

Video

Document

Top Task Types

Bounding Box

Classification

Segmentation

Object Detection

Question Answering

Fine-tuning

Freelancer Overview

I have direct, hands-on experience in end-to-end data labeling and AI training data preparation. As part of my M.Tech thesis at IIT Kharagpur, I manually labeled the Oxford-IIIT Pet Dataset from scratch, annotating 7,349 images across 37 pet breeds with segmentation masks, ensuring high-quality, annotation-ready data for training deep learning models like YOLOv5 and MobileNetV3. What sets me apart is my complete understanding of how data quality drives model performance. Beyond labeling, I built full image preprocessing pipelines (YCbCr conversion, DCT transformation, quantization, run-length encoding) and validated data at every stage before model training. With proficiency in Python, OpenCV, NumPy, and PyTorch, I bridge the gap between raw data and production-ready AI models, making me a strong fit for roles in data labeling, annotation pipelines, or AI training data preparation.

Education

Work History