Chirag Garg

Machine Learning & AI Researcher

Bridging theory and practice in machine learning with a focus on Computer vision and Representation learning.

Download CV

About Me

Chirag Garg

chirag.garg.5293@gmail.com

I'm a B.Tech student at National Institute of Technology, Raipur, majoring in Computer Science and Engineering with a focus on machine learning theory, representation learning, and computer vision.

My work revolves around understanding the core principles of Computer vision and representation learning. With a strong background in mathematics, I aim to bridge theory and practice to create AI models that are both powerful and interpretable. My vision is to develop algorithms that not only push the boundaries of what's possible in AI but also ensure these advancements are practical and aligned with real-world challenges.

ML Theory Deep Learning Optimization Computer Vision Representation Learning

Education

National Institute of Technology, Raipur

Bachelor of Technology in Computer Science and Engineering

2022-2026 | Raipur, India

CGPA: 8.23

Till 5th semester

Focusing on developing intelligent systems with an emphasis on machine learning theories and applications.

Experience

AI Research Intern

CiSTUP @ IISc Bengaluru | Bengaluru, India

May 2024 - July 2024
  • Worked on developing a Deterministic Autoencoder Model which is conceptually similar to VAE but free from variational inference.
  • Contributed to deriving an optimal constant stepsize for Adam under which it guarantees convergence.

Publications

Published

DIME: Deterministic Information Maximizing Autoencoder

ICLR 2025 Workshop

View Paper

Authors: Alokendu Mazumder, Chirag Garg, Tirthajit Baruah, Punit Rathore

This work proposes a novel approach to deterministic autoencoders that maximizes information content without relying on variational methods.

Under Review

A Theoretical and Empirical Study on the Convergence of Adam with Exact Constant Step Size in Non-Convex Settings

IEEE Transactions on Artificial Intelligence

Authors: Alokendu Mazumder, Rishabh Sabharwal, Bhartendu Kumar, Manan Tayal, Chirag Garg, Punit Rathore

This research investigates the convergence properties of Adam optimizer with constant step sizes in non-convex optimization landscapes.

Projects

SmartDoorGuard
Security System

SmartDoorGuard: Enhanced Door Access System

Advanced security system with dual verification integrating number plate recognition and face recognition.

Key Features

  • YOLOv8 for number plate recognition
  • Arcface for face recognition
  • MongoDB for efficient database management
Completed: March 2024 View Project
PotholeVision
Road Maintenance

PotholeVision: Smart Road Rehabilitation

Cutting-edge software for pothole detection, depth estimation, and 3D modeling for road maintenance.

Key Features

  • Real-time pothole detection using SOTA algorithms
  • Technologies: YOLOv8, Arcface, MongoDB, React, Firebase
Completed: Nov 2023 View Project

Skills

Programming

Python
C/C++

Technologies

Sklearn
TensorFlow
PyTorch
OpenCV

Mathematics

Linear Algebra
Probability
Statistics
Calculus

Concepts

Computer Vision
Deep Learning
Optimization
Machine Learning

Contact

Interested in collaborating on research or discussing machine learning concepts? I'm always open to connecting with fellow researchers and enthusiasts.