Osama Amjad

Lead AI @ AIO

Islāmābād, Pakistan

About

Hi, I'm Osama, an AI engineer specializing in developing cutting-edge AI-driven solutions for Original Equipment Manufacturers (OEMs) in the automotive industry. My innovative contributions have been instrumental in transforming the technology landscape for global brands, including Toyota, Haval, Changan, and Thal Engineering. Expertise: Advanced AI/ML model research, development, and deployment Pioneering research in autonomous systems tailored to challenging environments Leadership in mentoring teams, fostering innovation, and delivering impactful results Currently, I am a Senior AI Engineer at AIO, a U.S.-based company headquartered in San Jose, California. AIO offers the world's first all-in-one AI platform for restaurants, streamlining operations, marketing, accounting, staff, and inventory management through intelligent automation. I thrive at the intersection of innovation and execution, delivering solutions that not only push boundaries but also create tangible, real-world impact. Let's connect to explore the transformative potential of AI.

Experience

  • AIO (On-site)
    • Lead AI
      Feb 2026 - Present · 5 mos

    • Team Lead | Senior AI Engineer
      Jul 2025 - Feb 2026 · 8 mos

  • VisionRD (Islāmābād, Pakistan)
    • Head of AI and R&D
      Aug 2024 - May 2025 · 10 mos

      - Real-time video based action recognition system based on vFusedAct deployed at Thal Engineering, HAVAL Pakistan, KIA Lucky Motors for maximum efficiency of workers. - InspectionAI based on custom built inspection architectures deployed at RapidEV, producing 10-15 Million PCB’s per year. Each PCB get inspected by our system before going to market. - vRD-OCR, won the Most Innovative Project at Ready Tensor, Inc. Expo 2024. This same vRD-OCR, AI-driven document and quality inspection solution currently deployed at Toyota Pakistan

    • Senior Computer Vision Engineer | Team Lead | Vision Drive Platform
      May 2023 - Aug 2024 · 1 yr 4 mos

      - vFusedSeg3D: Won the Waymo Open Dataset Challenge 2024. Technical document accepted at CVPR 2024. - vONErd Model: Introduced the vONErd model, inspired by Tesla's innovative approaches. This model performs simultaneous tasks with exceptional accuracy: + 3D/2D Object Detection: Achieves an 80% mean average precision (mAP) for precise object identification. + Lane Detection: Sets a new standard with a 95.8% F1@IoU(0.5) for lane detection. + Segmentation: Excels in understanding and segmenting the environment. - PDrive20K: Pioneering Autonomous Driving Data Collection: Led the development of Pakistan's groundbreaking data collection efforts for autonomous driving, resulting in the creation of PDrive20K. This dataset is recognized for its exceptional data quality and diversity, and plays a pivotal role in advancing ADAS and AD technology. - Unparalleled Speed: The vONErd model boasts an industry-leading latency of 55ms, outperforming global ADAS solutions by 9%. Its real-time capabilities ensure reliability in challenging environments, such as Pakistan's diverse conditions. - Mentorship and Growth: Cultivated a culture of continuous learning and growth by providing guidance and mentorship to junior computer vision engineers, enabling them to excel and innovate in this dynamic field. - Cutting-Edge Research: Spearheaded the research and implementation of state-of-the-art computer vision algorithms and deep learning models tailored for ADAS applications, addressing the unique challenges posed by Pakistan's environmental conditions.

    • Computer Vision Engineer
      Jul 2022 - Aug 2024 · 2 yrs 2 mos

      - Successfully executed 4-week pilot with leading SUV OEM in Pakistan. Core Team Member of DriveAI Product. - Developed integrated model architecture for accurate 3D/2D object and lane line detection. - Achieved 50+ fps on semantic segmentation and lane line detection, and 3D object detection using CUDA programming and TensorRT deployment. - Built Front Collision Alert, Pedestrian Cross Alert, 360° Vehicle Monitor prototype for Pakistani environment. - Built SOTA model for Pakistani Road Sign Detection and Classification. - Proficient in Amazon SageMaker, EC2 for model training/deployment. Used Amazon S3 for efficient data handling and CVAT for annotations. - Worked on uncertainty estimation, scenario extraction for autonomous driving using Pakistani data. - Built AI model for millimeter-level defect detection in InspectionAI product. - Designed an approach to increase frame rate of high-resolution cameras in gphoto2. - Selected as top 15 startup in Pakistan, received grant from Innovator Seed Fund, World Bank, i2i Ventures, and HEC Pakistan. - Top 5 startup in Pakistan for Startup Germany, selected by German Emirati Joint Council for Industry & Commerce AHK.

  • IC Architecture Design Intern at National Institute of Electronics
    Jul 2020 - Dec 2020 · 6 mos

    Tasked with the IC design lab to understand the procedure of an integrated circuit design.Received training and certification of IC architecture design in Xilinx.