CV

PhD in Informatics (Mathematical Optimization)

• Doctoral research areas: mathematical optimization; optimization for machine learning; second-order and stochastic methods; randomized numerical linear algebra
• Thesis: The Utilization of Second-Order Information for Large-Scale Unconstrained Optimization Problems (PDF)
• Doctoral study funded by the Japan International Cooperation Agency (JICA)
• GPA: 3.7/4.0

Research Student, Mathematical Optimization

• System Optimization Laboratory

MSc, Mathematics

• Thesis: Hasse-Minkowski Principle for Quadratic Forms over Q
• Admitted via the Joint Admission Test (JAM), all-India rank 343

BSc, Mathematics

Machine Learning Research Engineer

• Project owner for three concurrent R&D lines in optimization and AI for science.
• Budget-aware Bayesian optimization for autonomous experimentation: a closed-loop harness in which every proposed experiment is priced, charged against a budget, and refused if unaffordable, so policy quality is always reported beside spend. Twelve proposal policies, three acquisition optimizers, and a one-command reproducibility check that re-derives every committed figure from committed data.
• Search heuristics for computer-aided synthesis planning: a learned cost-to-go estimate reused to reorder candidate reactions by remaining distance to available starting materials, improving solve rate and reducing search effort at unchanged per-step compute. First-author conference abstract.
• An installable, versioned software component publishing a data contract to a separate optimization engine, with determinism enforced as a test, module boundaries checked at process level, and continuous integration gating its own published result counts.
• Methodology across all three: pre-registered experiment designs, paired statistical tests with bootstrap confidence intervals, a stated detection floor beside every comparison, and negative results reported with their mechanism.

Machine Learning Research Intern

• Built a neural-guided bidirectional search method for double-ended synthesis planning, extending a published planner to handle convergent routes via a multi-tree forest and a proof-tree-complete meeting criterion.
• Implemented Monte Carlo Tree Search with UCT and PUCT variants and graph-based heuristics for retrosynthetic pathway search; benchmarked against established baselines over public chemical databases.
• Built multi-objective scoring for route selection across path length, cost and yield.

Machine Learning Research Associate

• Developed a Nystrom-based ADMM algorithm for federated learning, targeting the communication and memory cost of second-order information in distributed training.
• Built a Python framework for decentralised training across multiple nodes, with convergence monitoring and benchmarking.
• Co-PI; guided two M.Tech researchers. Delivered an invited talk on the method.

Research Assistant

• Developed the regularized Nystrom method for large-scale optimization, reducing per-iteration cost from O(nd^2 + d^3) to O(ndm + dm^2) with sketch size m of roughly 5-20% of the feature dimension, and established global convergence under standard assumptions.
• Benchmarked against randomized subspace Newton and Newton-sketch on high-dimensional classification.
• Applied curvature-aware optimization to brain-MRI tumour classification with transfer learning.

E-Commerce & Marketing Research Analyst (part-time)

• Held alongside doctoral research. Managed B2B client relationships and negotiations for international e-commerce, conducted competitive analysis for listings and pricing, and automated sales and inventory reporting.

Co-investigator, Encouragement Research Grant (1,000,000 JPY)

• Project: RL-BFGS for Big Data with Improved Global Convergence.
• Named co-investigator on an industry-funded grant as a second-year PhD student; responsible for numerical simulations and convergence analysis of the proposed regularized L-BFGS method.

JICA Scholarship for Doctoral Study

• Full funding for doctoral study at Kyoto University.

GSET - Gujarat State Eligibility Test

• Qualified for Assistant Professor eligibility in state universities.

Merit-cum-Means Scholarship

Joint Admission Test (JAM) - All-India Rank 343

Reviewer

Co-PI, Federated Learning project

• Guided two M.Tech researchers.

Vice President

• Core member of the Executive Committee, responsible for institutional and community partnerships, including academia-industry collaboration between the institute and organisations in Japan.

President (Treasurer 2019-2022)

• Led a volunteer team organising cultural events for 200+ participants, including visits by the Mayor of Kyoto, the Consulate General of India in Osaka-Kobe, and members of the Japanese House of Representatives.