Vismay Patel

Hi, I'm Vismay

Applied Scientist & ML Researcher

7+ years in Computer Vision, NLP, Deep Learning & Generative AI • IIT Madras Alumnus

Scroll

About Me

Vismay Patel

Applied Scientist

I specialize in architecting and deploying scalable machine learning solutions across Computer Vision, NLP, Deep Learning, and Generative AI. With over 7.5 years of experience and a Master's in Research from IIT Madras, I have a proven track record in technical leadership and translating cutting-edge research into production-ready solutions.

Currently at Amazon's Ads Trust team, I lead the development of multimodal content labeling systems, regulatory compliance pipelines, and billion-scale image retrieval platforms using VLMs, LLMs, and diffusion models.

I am also an avid photgrapher and highly interested in computational photography.

7+ Years Experience
4+ Publications

Work Experience

My professional journey in ML, AI, and software engineering.

January 2022 – Present

Amazon

Applied Scientist - Ads Trust Team

  • Led architecture for a multimodal content labeling system using VLMs achieving 90% coverage at 95% precision
  • Architected regulatory compliance system for pharmaceutical ads
  • Implemented real-time content moderation and billion-scale object-centric image retrieval platform
  • Developed synthetic data generation pipelines and Generative Image Models
April 2020 - December 2021

Verisk Analytics Inc.

Machine Learning Engineer - Global R&D Team

  • Developed enterprise-scale document information extraction and synthetic document generation pipeline
  • Architected pandemic modeling systems using agent-based network simulations
August 2018 - April 2020

CodeNation LLC.

Software Development Engineer

  • Deployed "ML Factory" - an AutoML platform for streamlined ML model training and deployment
  • Built "AutoFix" - AI-based code quality improvement using code-specific knowledge graphs
  • Developed time-series forecasting and anomaly detection models for multi-variate time-series
February 2018 – April 2018

Find Me A Shoe

Intern

  • Developed algorithm for foot measurement extraction from multi-view images using pose estimation techniques

Education

Academic background and research experience.

🎓

Indian Institute of Technology Madras

2014 – 2018

CGPA: 8.8/10

M.S. in Computer Science & Engineering

Specialization in Computer Vision and Deep Learning. Thesis: "Data Augmentation using Part-based Deformations of Shapes." Also worked on Pedestrian Detection, Sketch-based Image Retrieval, Bi-modal Personality Analysis, and Image & Video Inpainting.

📚

Institute of Technology, Nirma University

2010 – 2014

CGPA: 7.76/10

B.Tech. in Information Technology

Strong foundation in algorithms, data structures, and software engineering fundamentals.

Projects & Publications

Research papers, competition wins, and featured work.

Data Augmentation with Part Analysis
WACV 2019
Matlab Torch

Data Augmentation using Part-based Deformations of Shapes

M.S. thesis research - using part analysis of shapes extracted from available images to augment labeled data, achieving improved CNN classifier performance for shape classification.

Personality Analysis from Videos
ECCV 2016 Workshop
Lua Torch

Personality Analysis from Interview Videos

Deep Learning solution for the "First Impressions" challenge by Chalearn Looking at People. Employed a novel training technique for videos - scored 2nd rank in the competition.

Video & Image Inpainting
ECCV 2018 Workshop
Python PyTorch GANs

Video and Image Inpainting

GAN-based techniques for caption removal from videos and images. Includes joint caption detection and inpainting - scored 3rd rank in the Chalearn competition.

Technical Skills

Technologies, frameworks, and research domains.

💻 Languages and Frameworks

Python Java Lua C++ PyTorch Transformers Sklearn JAX TensorFlow OpenCV

🔬 Domains

Computer Vision NLP Deep Learning Generative AI Multimodal Models Explainable AI Active Learning

☁️ MLOps / Cloud

AWS Neo4j Kubernetes Docker Kubeflow Kubeflow VLLM SGLang

Get in Touch

Have a question or want to collaborate? Let's talk!