CV

PhD in Mathematical Optimization

• GPA: 3.7/4.0
• Sponsored by Japan International Cooperation Agency (JICA)
• Thesis: The Utilization of Second-Order Information for Large-Scale Unconstrained Optimization Problems

Research Student in Mathematical Optimization

Master of Science in Mathematics

• Hasse-Minkowski's Principle for Quadratic forms over Q, Arithmetic, Number theory

Bachelor of Science in Mathematics

Research Associate

• Developing Nyström-based ADMM Optimization algorithm for Federated Learning to improve computational efficiency.
• Investing distributed convex and non-convex optimization techniques for large scale decentralized training.
• Analyzing convergence and reducing communication overhead using quantization using Python.
• Preparing research paper for submission to top-tier ML/Optimization Journal/conference.

E-Commerce & Back Office (Part time)

• Conducted in-depth market research and trend analysis for e-commerce platforms, optimizing auto parts listings to boost global sales on eBay and Amazon.
• Managed client relations while facilitating negotiations, and ensuring customer satisfaction, contributing to securing major orders and achieving consistent sales growth.
• Streamlined online sales operations, processes and enhanced productivity through process optimization using Excel and MS Office Suite, contributing consistent revenue growth and securing major international orders.
• Identified key sales trends and provided actionable insights, enhancing strategic decision-making and improving sales targets.

Research Assistant

• Pioneered the development of randomized algorithms, including the Regularized Nyström method, reducing computational CPU time by 50-70% for large-scale and high-dimensional optimization problems.
• Developed a computationally efficient regularized Nyström method that reduces complexity from O(nd² + d³) (Newton's method) to O(ndm + dm²), where m is only 5-20% of d (dimension/features), significantly enhancing scalability for high-dimensional datasets.
• Conducted medical image analysis on brain MRI data using deep learning and transfer learning techniques, advancing tumor detection accuracy by 86%.
• Built and validated machine learning models, integrating logistic regression, linear regression, SVM, gradient-based methods, and stochastic optimization for high-dimensional datasets.
• Presented research findings at esteemed international conferences, showcasing innovation in scalable matrix approximations, optimization methods for convex and non-convex functions.

Online Expert of Mathematics

• Created a series of in-depth tutorials covering complex mathematical concepts.
• Provided accurate, expert-level solutions to complex graduate-level mathematics problems, demonstrating strong analytical, problem-solving skills, and enhancing academic success for global learners.

KIOXIA

• Secured ¥1 million funding to develop scalable stochastic optimization algorithms, reducing computational complexity by a factor of d (# of parameters) while maintaining accuracy.
• Enhanced performance of regularization techniques for large-scale datasets through numerical experiments on benchmark datasets, improving optimization behavior analysis.
• Designed and implemented an innovative SVRG-RL-BFGS optimization algorithm, reducing computational cost for large-scale machine learning problems.
• Applied dimensionality reduction techniques like random projection on high-dimensional datasets, improving computational efficiency and classification accuracy.
• Investigated high-dimensional optimization challenges, combining quasi-Newton methods with variance reduction techniques to develop scalable solutions for empirical risk minimization problems.

President

Treasurer

Responsible for leading a team of 5 students

• Ensuring successful project completion.

• Japan International Cooperation Agency (JICA) Scholarship for PhD at Kyoto University

• Cleared GSET exam for Assistant Professor in Govt. Universities, Gujarat, India

• Got 1 million Japanese Yen funding for one year industrial project at KIOXIA

• Awarded MCM Scholarship during master's program

• Cleared JAM with All India rank 343rd

• Cleared All India Engineering Entrance Exam

Reducing Variance of Stochastic Gradient using Barzilai-Borwein method as second-order information

• Nagaoka Institute of Technology, Niigata, Japan

A Stochastic variance reduced gradient using second order information

• Waseda University, Tokyo, Japan

Nys-Newton: Nyström approximated curvature information for convex optimization

Nys-Transfer: Nyström approximated Newton-sketch for Fine-tuning the Deep Nets for Brain MRI

• Okinawa, Japan

Regularized L-BFGS method for Large Scale Unconstrained Optimization

• Berlin - Germany

Technical Review of Scientific Articles