Santosh Shriyan - Resume
Santosh Shriyan - Resume
Experienced Computer Vision Engineer with a strong background in Python, AI/ML solution design, and
cloud-based process automation. Proficient in developing deep learning algorithms and integrating
AWS for seamless operations. Skilled in technical research and communication, ready to contribute to
the organization's Computer Vision needs.
WORK EXPERIENCE
peAR Technologies, Mumbai
Computer Vision/AI Engineer March 2021 - April 2025
• Designed and Developed smart phone video to 3D system replacing older DSLR capture and manual
processing.
• Trained, fine tuned and deployed Super-Resolution, YOLOv8 based detection and segmentation
among other Deep Learning algorithms to boost accuracy and performance.
• Integrated AWS EC2 and S3 for cloud based process automation and custom autoscaling.
• Responsible for Technical Research for Computer Vision and 3D asset creation.
Cipla, Mumbai
Trainee Engineer (Electrical) July 2013 - December 2013
• Handled various pharmaceutical machineries according to their respective batches.
• Maintained calibration of various instruments.
• Maintained batch historic records.
SKILLS
• Python, OpenCV, Image processing
• AWS, Docker, Linux Bash
• Yolo.v8, Super-Resolution
• Pytorch, Data Analysis
• Electronics/Embedded hardware
• Good communication skills
PROJECTS
Mobile Video to 3D System - peAR Technologies April 2025
• This is the flagship product of peAR Technologies Pvt Ltd.
• The project was to move from DSLR (RAW) images based capture to a mobile video based 3D asset
creation.
• I was responsible for the complete system, from Research -> Development -> Testing -> Debugging
-> Deployment (AWS) -> Feature Request implementation for its complete life cycle.
• There were multiple iterations made to the product including but not limited to Super-Resolution,
Object Detection, Object segmentation, Blur Detection, Debluring, Object tracking, Blender(3D) CLI
automation, Server Load Balancing (peAR), etc
Vision based Vehicle Detection and Tracking System April 2018
• As part of pre-processing background subtraction was used to limit computation for feature
detection and tracking.
• SURF was used as a primary feature detection technique, edge detection was also used in times of
lower feature points (to avoid inconsistent tracking).
• The detected feature points were then clustered using DBSCAN algorithm and parameterised to
avoid background movement (outlier feature points)
Certifications
Machine Learning with Python: IBM
Credential ID: RTJJFXGTYS6Z
Python for Data Science, AI & Development: IBM
Credential ID: FRGLW8RDZBEF
AWS Cloudfront: Serve content from multiple S3 buckets: Coursera
Credential ID: HFTZESLFRKLC
Introduction to Git and GitHub: Google
Credential ID: U7BEN9TMHNEK
Industry 4.0: Globsyn Business School Online
Credential ID: GBSO-002-2020-24424