Roshan Rayala Bhaskar

Master’s student in Digital Engineering with a strong foundation in Machine Learning, Deep Learning, and 3D Computer Vision. Experienced in developing AI-driven solutions, including generative models, view synthesis, and automation for CAD workflows. Passionate about research, with hands-on experience in Python, PyTorch, and computer vision frameworks. Adept at bridging engineering and AI to drive innovation in digital design and automation.

• Fine-tuning Large Vision-Language models for CAD data extraction from engineering drawings | Dec 2024 – Present
  Evaluating Vision-Language Models to extract Welding Data from engineering CAD drawings, Fine-Tuning for domain-specific use-case, and creating a streamlined framework suitable for an SME
• 2D Image to 3D point cloud – Monocular Depth Estimation | Jan 2024 – March 2024
  Developed a 2D to 3D reconstruction pipeline using PyTorch and Transformers with depth estimation, point cloud generation, and 3D mesh reconstruction
• Unity VR Training Application for Neurosurgeon Training | Dec 2023 – July 2024
  Develop Unity VR app aimed at training Neurosurgeon’s Visio-spatial ability in analysing DSA angiography images, incorporating custom shaders, 3D models, and gamification scripts using C#

Diploma in Artificial Intelligence & Machine Learning, University of Hyderabad - Online | 08/01/21 - 08/31/22

Acquired essential skills in Foundational Mathematics, Essentials of AI, Data Analysis and Visualization, Machine Learning, and Deep Learning, leading to the completion of multiple machine learning case studies and a thesis project addressing a challenging prediction problem

• Introduction to Deep Learning: Completed an in-depth Deep Learning course covering MLPs, CNNs, RNNs, Transformers, and generative models
• Learning Generative Models: Implemented deep learning concepts using TensorFlow and PyTorch in real-world scenarios - VAEs, GANs, Diffusion
• Advanced Topics in 3D Computer Vision: Presented seminars on topics like implicit decoder neural networks for generative shape modeling and Decoder only - Transformers for generative mesh modeling, and Novel View Synthesis
• Research Interests: Explore video generation models for view synthesis and 3D generation, Optimize Product Design workflows

• English: C1
• German: B1

• Date of Birth: 13/AUG/1994
• Nationality: Indian