Harsh Pateriya
Manager (Research & Technology Application) Data Science & Artificial Intelligence Department @TATA STEEL TECHNICAL SERVICES LTD
Signup · Get unlimited contacts
WORK HISTORY
Manager (Research & Technology Application) Data Science & Artificial Intelligence Department @TATA STEEL TECHNICAL SERVICES LTD
Kolkata, IN
Computer Vision & Deep Learning for Industrial Systems- Designed and deployed CNN-based computer vision systems for automated surface defect classification in rolled steel manufacturing- Developed multi-class defect detection pipelines incorporating domain-specific augmentation, class imbalance handling, and deployment-aware optimization- Applied transfer learning and progressive CNN architectures to improve generalization across surface textures and process conditions- Validated models under production-like constraints with emphasis on reliability, inference efficiency, and interpretability for manufacturing decision support.Large-Scale Vision & Spatial Intelligence- Built scalable semantic segmentation pipelines for high-resolution satellite imagery using DeepLabv3+, UNet, LinkNet, and YOLO- Designed automated data ingestion and preprocessing pipelines to support robust model training on heterogeneous, noisy datasets- Translated vision outputs into actionable industrial insights for infrastructure and market analysis.
EDUCATION
Indian Institute of Technology, Kharagpur
Master of Technology , Manufacturing Science & Engineering
Indian Institute of Technology, Kharagpur
Bachelor of Technology, Mechanical Engineering
Campion School Bhopal
High School, Mathematics, Physics, Chemistry
ABOUT HARSH PATERIYA
I work at the intersection of computer vision, applied AI, and real-world industrial systems, with a primary focus on designing deep learning models that function reliably beyond laboratory conditions.My current role as Manager – Data Science & Artificial Intelligence at Tata Steel involves building and deploying computer vision pipelines for manufacturing quality inspection and large-scale industrial decision support. My work centers on convolutional neural networks, transfer learning, semantic segmentation, and deployment-aware modeling, with explicit attention to robustness, generalization, and interpretability under real production constraints.My formal training is in Manufacturing Science & Engineering and Mechanical Engineering from Indian Institute of Technology Kharagpur, which provides the domain grounding that informs how I design AI systems for manufacturing environments. This background allows me to frame vision models not only as pattern recognizers, but as tools for understanding process behavior, defect formation, and production variability.Alongside industry work, I maintain active research engagement in deep learning and computer vision. I have published in PLOS ONE on spatiotemporal modeling of satellite imagery using ConvLSTM–CNN architectures and have presented research on multimodal and meta-learning approaches for image-based medical diagnostics. Across domains, my research focus remains consistent: building learning systems that generalize across data shifts and operate reliably in real-world conditions.I am now seeking PhD training focused on computer vision and AI methods for manufacturing science, particularly in smart manufacturing and Industry 4.0. My goal is to advance vision-based learning systems that improve manufacturing quality, process understanding, and decision-making at scale.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.