I am a third-year Computer Science undergraduate with focused research experience in Computer Vision and Machine Learning. My primary research interest is robust visual recognition under distribution shift, specifically how classical feature engineering (HOG, SIFT) and deep convolutional architectures generalize as scene complexity grows. I benchmarked six classifier families on Fashion-MNIST and CIFAR-10, performed 5-fold cross-validation with full hyperparameter search, and published findings in IEEE format. I also deployed a production Vision Transformer classification system and built an LLM knowledge-distillation pipeline to reduce dataset size by 40% while preserving label quality. I am seeking a supervised summer research placement (June–August 2026) in CV or ML to deepen expertise and contribute to ongoing group research (e.g., 3D scene understanding or self-supervised learning).

Mohammad Usman Ahmad

I am a third-year Computer Science undergraduate with focused research experience in Computer Vision and Machine Learning. My primary research interest is robust visual recognition under distribution shift, specifically how classical feature engineering (HOG, SIFT) and deep convolutional architectures generalize as scene complexity grows. I benchmarked six classifier families on Fashion-MNIST and CIFAR-10, performed 5-fold cross-validation with full hyperparameter search, and published findings in IEEE format. I also deployed a production Vision Transformer classification system and built an LLM knowledge-distillation pipeline to reduce dataset size by 40% while preserving label quality. I am seeking a supervised summer research placement (June–August 2026) in CV or ML to deepen expertise and contribute to ongoing group research (e.g., 3D scene understanding or self-supervised learning).

Available to hire

I am a third-year Computer Science undergraduate with focused research experience in Computer Vision and Machine Learning. My primary research interest is robust visual recognition under distribution shift, specifically how classical feature engineering (HOG, SIFT) and deep convolutional architectures generalize as scene complexity grows.

I benchmarked six classifier families on Fashion-MNIST and CIFAR-10, performed 5-fold cross-validation with full hyperparameter search, and published findings in IEEE format. I also deployed a production Vision Transformer classification system and built an LLM knowledge-distillation pipeline to reduce dataset size by 40% while preserving label quality. I am seeking a supervised summer research placement (June–August 2026) in CV or ML to deepen expertise and contribute to ongoing group research (e.g., 3D scene understanding or self-supervised learning).

See more

Experience Level

Expert
Expert
Expert
Expert
Expert
Intermediate
Intermediate
Intermediate
See more

Work Experience

AI/ML Research Intern at Arch Technologies
February 1, 2026 - April 1, 2026
Deployed a multi-modal Vision Transformer pipeline integrating three Hugging Face ViT models (ImageNet1K general classification, 7-class facial emotion recognition, CheXpert chest X-ray pathology detection) with a FastAPI backend and React/TypeScript frontend; live at smart-detection.vercel.app. Built an LLM dataset distillation pipeline applying knowledge distillation, heuristic filtering, and perplexity-based quality scoring reducing dataset size by 40% while preserving label quality for downstream fine-tuning tasks. Designed and published dataaudit to PyPI: an open-source Python library that replaces 1015 manual EDA inspection steps with a single function call for ML practitioners.

Education

BS Computer Science at UIIT, PMAS Arid Agriculture University
September 1, 2023 - May 1, 2027
AI/ML/DL Diploma at Air University
February 1, 2026 - May 1, 2026

Qualifications

NAVTTC Certification
January 11, 2030 - June 10, 2026

Industry Experience

Software & Internet