MLBench Pvt LTD

MLBench is an AI and computer vision development shop based in Pakistan, founded in 2017. The company has delivered 100+ projects for 50+ clients over 15 years, and employs 25+ people across engineering, design, QA, and business development roles. Led by CEO Nabeel Hassan and Co-Founder Qazi Ammar Arshad, with CTO Waqas Sultani (PhD.), MLBench builds production-grade AI systems, document processing pipelines, and mobile applications for clients across the UAE, USA, Europe, Australia, and Asia. They work with startups and established companies in automotive, healthcare, fintech, surveillance, and industrial automation.

Services and capabilities

MLBench's core offering is custom AI and machine learning development. They build computer vision systems (object detection, image segmentation, real-time tracking), deploy large language models and vision-language models for document understanding, and handle NLP tasks like entity recognition and sentiment analysis. On the application side, they develop native iOS and Android apps, cross-platform solutions using React Native and Flutter, and enterprise dashboards and APIs.

Their document AI practice explicitly includes automated processing of MPG and STNK documents with structured data extraction and validation. They work with OCR, cell segmentation for medical imaging, anomaly detection systems, and fitness AI (analyzing barbell lift form from video). Their tech stack runs across TensorFlow, PyTorch, OpenCV, Scikit-learn, Keras, YOLO, GPT, LangChain, and Transformers on the ML side, and Swift, Kotlin, React, Node.js, Django, Flask on the application side. Cloud deployment is handled on AWS and Google Cloud with Docker and Kubernetes.

Notable work

They showcase concrete projects with measurable outcomes. A padel sports analytics platform delivers real-time player tracking, shot classification, and tactical insights. A T-shirt counting system handles large-scale inventory matching. Passenger tracking uses overhead video in airport queues with stated high accuracy. Medical work includes brain MRI segmentation and malaria cell detection from microscopy images. Computer vision projects span drone detection (two-stage system with segmentation and attention), theft detection using pose estimation and keypoint tracking, and a UV/hygiene system combining multi-person tracking with hand detection for kitchen workflow optimization. They also built an ML rice classifier for mobile that predicts rice genus and quality from photos, and "Pure Strength AI Trainer," an app that analyzes barbell lift video to provide form feedback, velocity, and range-of-motion data.

How they work

MLBench's process follows five stages: comprehensive analysis and planning with strategic insights, ideation and strategy development tailored to client goals, iterative design and prototyping based on feedback, development and execution using optimized ML frameworks and pipelines, and deployment with continuous performance evaluation. They assign dedicated ML teams to clients and emphasize data security with enterprise-grade protections at each pipeline stage. Pricing and engagement length are not specified in the source material, but they offer free consultations.

Team and credentials

The leadership includes CEO Nabeel Hassan, Co-Founder Qazi Ammar Arshad, and CTO Waqas Sultani (PhD.). The 25+ person team is distributed across software engineering (Principal, Senior, Associate, and Junior roles), AI/ML specialists, UI/UX designers, QA engineers, business development, and HR. No ISO, SOC 2, or partner certifications are mentioned. One client testimonial is attributed to a "Curry" startup founder who calls the engagement a "game-changer" for app development, though no specific outcome metrics are provided.