Why Global LLM Leaderboards Are Misleading: Small Portfolios for Heterogeneous Supervised ML
Under submission, with Ayela Chughtai, Bhargavi Lanka, Swati Gupta.
CV
Education
- B.Tech. in Computer Science and Engineering, Indian Institute of Technology, 2015-2019
- Visiting Student at Sloan School, Massachusetts Institute of Technology, Fall 2024
- Ph.D. in Algorithms Combinatorics and Optimization, Georgia Tech, Spring 2021-2025
Publications
All
Many Preferences, Few Policies: Towards Scalable Language Model Personalization
Under submission, with Cheol Woo Kim, Roozbeh Nahavandi, Andrew Perrault, Milind Tambe, Swati Gupta.
Stochastic Function Certification with Correlations
Under submission, with Rohan Ghuge and Mohit Singh.
Navigating the Social Welfare Frontier: Portfolios for Multi-objective Reinforcement Learning
ICML, with Cheol Woo Kim, Shresth Verma, Madeleine Pollack, Lingkai Kong, Milind Tambe, Swati Gupta.
Balancing Notions of Equity: Trade-offs Between Fair Portfolio Sizes and Achievable Guarantees
SODA, with Swati Gupta and Mohit Singh.
Promise of Graph Sparsification and Decomposition for Noise Reduction in QAOA: Analysis for Trapped-Ion Compilations
Under revision at Quantum, with Philip C. Lotshaw, Greg Mohler, and Swati Gupta.
On Disjunctive Rado Numbers for Some Sets of Equations
The Electronic Journal of Combinatorics, with A. Dileep and Amitabha Tripathi.
Warm-Started QAOA with Custom Mixers Provably Converges and Computationally Beats Goemans-Williamson’s Max-Cut at Low Circuit Depths
Quantum, with Reuben Tate, Bryan Gard, Greg Mohler, and Swati Gupta.
Which $L_p$ norm is the fairest? Approximations for fair facility location across all “$p$”
Economics and Computation (EC), with Swati Gupta and Mohit Singh.
Multi Purpose Routing: New Perspectives and Approximation Algorithms
arXiv, with Majid Farhadi, Prasad Tetali, and Alejandro Toriello.
Generating Target Graph Couplings for QAOA from Native Quantum Hardware Couplings
Physical Review A, with Joel Rajakumar, Bryan Gard, Swati Gupta, and Creston D. Herold.
New Proofs for the Disjunctive Rado Number of the Equations $x_1 - x_2 = a$ and $x_1 - x_2 = b$
Graphs and Combinatorics, with A. Dileep and Amitabha Tripathi.
Reusing Combinatorial Structure: Faster Iterative Projections over Submodular Base Polytopes
NeurIPS, with Hassan Mortagy and Swati Gupta.
Exact and Approximate Results on the Least Size of a Graph with a Given Degree Set
Discrete Applied Mathematics, with Aditya Sahdev and Amitabha Tripathi.
Why Global LLM Leaderboards Are Misleading: Small Portfolios for Heterogeneous Supervised ML
Jai Moondra, Ayela Chughtai, Bhargavi Lanka, Swati Gupta. (2026). "Why Global LLM Leaderboards Are Misleading: Small Portfolios for Heterogeneous Supervised ML." In arXiv.
)
Many Preferences, Few Policies: Towards Scalable Language Model Personalization
Cheol Woo Kim, Jai Moondra, Roozbeh Nahavandi, Andrew Perrault, Milind Tambe, Swati Gupta. (2026). "Many Preferences, Few Policies: Towards Scalable Language Model Personalization." In arXiv.
)
Stochastic Function Certification with Correlations
Rohan Ghuge, Jai Moondra, Mohit Singh. (April 2026). " Stochastic Function Certification with Correlations." arXiv preprint arXiv:2604.02611
)
Navigating the Social Welfare Frontier: Portfolios for Multi-objective Reinforcement Learning
Cheol Woo Kim, Jai Moondra, Shresth Verma, Madeleine Pollack, Lingkai Kong, Milind Tambe, Swati Gupta. (2025). "Navigating the Social Welfare Frontier: Portfolios for Multi-objective Reinforcement Learning." In arXiv.
)
Balancing Notions of Equity: Trade-offs Between Fair Portfolio Sizes and Achievable Guarantees
Swati Gupta, Jai Moondra, Mohit Singh. Symposium on Discrete Algorithms (Accepted, October 2024). "Balancing Notions of Equity: Trade-offs Between Fair Portfolio Sizes and Achievable Guarantees"
)
Promise of Graph Sparsification and Decomposition for Noise Reduction in QAOA: Analysis for Trapped-Ion Compilations
Jai Moondra, Philip C. Lotshaw, Greg Mohler, Swati Gupta. (June 2024). "Promise of Graph Sparsification and Decomposition for Noise Reduction in QAOA: Analysis for Trapped-Ion Compilations ." In arXiv
)
On Disjunctive Rado Numbers for Some Sets of Equations
A. Dileep, Jai Moondra, Amitabha Tripathi. (2024). "On Disjunctive Rado Numbers for Some Sets of Equations." The Electronic Journal of Combinatorics 31:1.
)
Warm-Started QAOA with Custom Mixers Provably Converges and Computationally Beats Goemans-Williamson’s Max-Cut at Low Circuit Depths
Reuben Tate, Jai Moondra, Bryan Gard, Greg Mohler, Swati Gupta. (September 2023). "Warm-Started QAOA with Custom Mixers Provably Converges and Computationally Beats Goemans-Williamson’s Max-Cut at Low Circuit Depths"
)
Which $L_p$ norm is the fairest? Approximations for fair facility location across all “$p$”
Swati Gupta, Jai Moondra, Mohit Singh. (July 2023). " Which $L_p$ norm is the fairest? Approximations for fair facility location across all "$p$"." In Proceedings of the 24th ACM Conference on Economics and Computation (EC) 2023
)
Multi Purpose Routing: New Perspectives and Approximation Algorithms
Majid Farhadi, Jai Moondra, Prasad Tetali, Alejandro Toriello. (2022). "Multi Purpose Routing: New Perspectives and Approximation Algorithms." arXiv.
)
Generating Target Graph Couplings for QAOA from Native Quantum Hardware Couplings
Joel Rajakumar, Jai Moondra, Bryan Gard, Swati Gupta, Creston D. Herold. (August 2022). "Generating Target Graph Couplings for QAOA from Native Quantum Hardware Couplings." Physical Review A 106.2
)
New Proofs for the Disjunctive Rado Number of the Equations $x_1 - x_2 = a$ and $x_1 - x_2 = b$
A. Dileep, Jai Moondra, Amitabha Tripathi. (2022). "New Proofs for the Disjunctive Rado Number of the Equations $x_1 - x_2 = a$ and $x_1 - x_2 = b$." Graphs and Combinatorics 38:38.
)
Reusing Combinatorial Structure: Faster Iterative Projections over Submodular Base Polytopes
Jai Moondra, Hassan Mortagy, Swati Gupta. (2021). "Reusing Combinatorial Structure: Faster Iterative Projections over Submodular Base Polytopes." In proceedings of the 34th conference on Neural Information Processing Systems (NeurIPS).
)
Exact and Approximate Results on the Least Size of a Graph with a Given Degree Set
Jai Moondra, Aditya Sahdev, Amitabha Tripathi. (2023). "Exact and approximate results on the least size of a graph with a given degree set." arXiv.
)
Other work experience
- Applied Science Intern at Amazon, Summer 2024
- Worked on a project to integrate simulation-based predictions with machine learning predictions.
Paper reviews for conferences and journals including FOCS, SODA, ITCS, IPCO, APPROX, ACDA, Math Programming, Operations Research, and SIDMA.
Session co-chair on Discrete Optimization with Swati Gupta at INFORMS 2023.
Research Assistant at Georgia Institute of Technology (January 2021 - present)
- Research Assistant at Duke University Computer Science Department, Summer 2018
- Advisors: Dr. Debmalya Panigrahi, Dr. Rong Ge
Talks
Bridging Theory and Practice for Balancing Notions of Equity with Small Portfolios at INFORMS 2024
Portfolio Approximations and Fairness in Combinatorial Optimization at INFORMS 2023
Which $L_p$ norm is the fairest? Approximations for fair facility location across all `$p$’ at EC 2023
Fairness Objectives in Facility Location Problems at INFORMS 2022
Academic awards and fellowships
Poster on Reusing Combinatorial Structure: Faster Iterative Projections over Submodular Base Polytopes received honorable mention for best poster at Mixed Integer Programming (MIP) 2022; joint work with Hassan Mortagy and Swati Gupta.
ARC-ACO fellowship at Georgia Tech for Spring 2023: this is jointly awarded by the Algorithms and Randomness Center and the Algorithms, Combinatorics, and Optimization program at Georgia Tech each year.