Open to opportunities

Hammad Ali Haider

@hammadalihaider

Message

Computer vision researcher applying continual learning and LoRA to detect anomalies.

China

Message

What I'm looking for

I’m looking for an ML/Computer Vision role where I can ship continual-learning anomaly detection with parameter-efficient fine-tuning on vision foundation models—building strong evaluation pipelines, iterating fast, and turning research into production-ready systems.

I’m a Computer Science master’s researcher focused on visual anomaly detection, continual learning, and parameter-efficient fine-tuning of vision foundation models. My work is hands-on: I’ve reproduced UniVAD on MVTec-AD, MVTec-LOCO, and VisA across 15+ categories, and implemented SplitLoRA-style continual learning experiments on standard CL benchmarks.

I build reliable training and evaluation pipelines around practical metrics like image-level AUROC, pixel-level AUPRO, and Backward Transfer (BWT). I also bring production-minded ML experience—fine-tuning BERT for sentiment and text classification at Bytewise, and shipping a full-stack AI web app (Next.js + Flask + MongoDB) that unified video-script preprocessing workflows from preprocessing to inference with end-to-end CI/CD.

Experience

Work history, roles, and key accomplishments

Current

Research Assistant

Current

Xinjiang University

Sep 2025 - Present (9 months)

Reproduced UniVAD (few-shot VAD) on MVTec-AD, MVTec-LOCO, and VisA across 15+ categories and implemented SplitLoRA for continual learning. Built LoRA-based, domain-adaptive fine-tuning pipelines on DINOv2/CLIP and an evaluation stack using image AUROC, pixel AUPRO, and Backward Transfer (BWT) metrics.

LoRAs SplitLoRA DINOv2 AUROC Backward Transfer (BWT)PyTorch DinoV2

Operations & IT Solutions Analyst

Punjab Information Technology Board (PITB)

Mar 2025 - Jun 2025 (3 months)

Delivered a working MVP for a clean-energy fintech product over a 4-month government-backed incubation and pitched it to a live investor panel. Conducted market validation, customer discovery interviews, and competitive analysis to iterate product positioning and unit economics over 4+ cycles.

Product Validation Customer Discovery Competitive Analysis Go To Market Strategy Product Positioning Unit Economics Stakeholder Management Fintech Clean Energy Domain

NLP Engineer (Fellowship)

Bytewise Limited

Jun 2024 - Aug 2024 (2 months)

Fine-tuned BERT for sentiment analysis and text classification, building end-to-end evaluation pipelines with F1, ROC-AUC, precision/recall, and confusion-matrix tracking. Reproduced 3+ state-of-the-art NLP papers over a 12-week fellowship and implemented reproducible preprocessing-to-inference workflows using Hugging Face Transformers and TensorFlow.

Sentiment Analysis TensorFlow Evaluation Metrics (F1 ROC AUC)Tokenization Cross Validation Confusion Matrix

Education

Degrees, certifications, and relevant coursework

Xinjiang University

Master of Science, Computer Science and Technology

2025 -

Activities and societies: Research direction: Visual Anomaly Detection; Continual Learning; Parameter-Efficient Fine-Tuning; Vision Foundation Models.

Research-focused Master of Science in Computer Science and Technology working on visual anomaly detection, continual learning, and parameter-efficient fine-tuning for vision foundation models.

COMSATS University Islamabad

Bachelor of Science, Computer Science

2021 - 2025

Grade: GPA: 4.0 / 4.0 (final-year project; top of cohort)

Activities and societies: Relevant coursework: Artificial Intelligence, Machine Learning, Natural Language Processing, Statistics & Probability Theory, Database Systems.

Completed a Bachelor of Science in Computer Science, achieving a final-year GPA of 4.0/4.0 and coursework covering AI, machine learning, NLP, statistics & probability, and database systems.