Zainab Rizwan

AI & Machine Learning Engineer with 2+ years of experience building and deploying production-grade ML systems across computer vision, OCR, generative AI, and LLM integration. Experienced across the full ML lifecycle; dataset annotation, model training, and cloud deployment achieving up to 95% accuracy on enterprise-grade benchmarks. Skilled in MLOps practices including FastAPI model serving, ONNX/TensorRT inference optimization, AWS Bedrock LLM integration, and vector databases, with applications spanning aerial object recognition, behavioral analytics, and document intelligence.

Computer Vision / AI Engineer at Neuralogic - B2B via CCRIPT Agency (2026-01 – Present)

• Design and deploy production-ready AI/ML pipelines for Neuralogic's enterprise clients, spanning object detection, OCR, segmentation, and vision-language model (VLM) tasks, achieving 93–95%+ accuracy benchmarks.
• Deploy CV and LLM pipelines on AWS EC2 GPU instances; reduced inference latency by optimizing pipelines with ONNX and TensorRT. Build and expose model-serving endpoints via FastAPI for seamless client integration.
• Currently developing an AI-powered CAD drawing analysis pipeline for automated construction takeoff estimation, aligned with US building standards.

Freelance Computer Vision Engineer at Upwork (2025-07 – 2025-12)

• Delivered a custom real-time detection and tracking system for an international client, using a fine-tuned feature-based identity descriptor to maintain consistent multi-object tracking across occlusion and re-entry events.
• Owned the project end-to-end; dataset annotation, model training, tracking pipeline development, and final delivery as sole engineer on a client-facing freelance engagement.

CV Research Assistant at Swarm Robotics Lab (SRL NCRA) (2024-06 – 2025-06)

• Researched and developed surveillance-oriented CV systems, evaluating state-of-the-art architectures including CNNs, YOLO variants, DETR, and lightweight models against real-time performance benchmarks.
• Designed and maintained scalable dataset pipelines supporting COCO, JSON, YOLO, and CSV annotation formats, reducing preprocessing overhead by 30% through automated formatting workflows.

Machine Learning Intern at CodeAlpha & Digital Empowerment Network (2024-07 – 2024-09)

• Built and benchmarked CNN-based classification models, improving training efficiency by 25% through systematic hyperparameter tuning across supervised learning tasks.
• Gained applied experience in model evaluation, data preprocessing, and experiment tracking across supervised learning tasks.

Bachelor of Science in Computer Engineering – University of Engineering & Technology (UET)

Pre-Engineering (FSc) – The Scholars Science College