News

  • Jul 2026Workshop Mixed-Policy GRPO for Text-to-SQL with Off-Policy Data Generation accepted for Workshop on Structured Understanding, Retrieval, and Generation in the LLM Era (SURGeLLM 2026).
    M. Sterbentz, M. Glass, N. H. Pham, D. Subramanian, K. J. Hammond.
  • Jun 2026Demo Text-to-SQL Evaluation Toolkit accepted for 52nd International Conference on Very Large Data Bases (VLDB 2026 Demo Track).
    Oktie Hassanzadeh (IBM Research)*; Yotam Perlitz (IBM); Nhan Pham (IBM); Tanvi Kaple (IBM); Karolina Źróbek (IBM); Long Vu (IBM); Michael Glass (IBM); Dharmashankar Subramanian (IBM); Mohammadreza Pourreza (University of Alberta); Davood Rafiei (University of Alberta)
  • Apr 2026Patent Application Generating Dynamic Few-Shot Examples for Enterprise Text-to-SQL Tasks
    L. H. Vu, D. Snoddy, T. R. Kaple, T. R. Dinger, N. H. Pham, M. R. Glass, D. Subramanian, A. W. Hagleitner.
  • Feb 2026Patent Granted System and Method for Combining Data Selection and Reward Function for Tuning LLMs using Reinforcement Learning. Patent US12566929B2.
    L. Vu, N. H. Pham, D. Subramanian, T. Mummert.
  • Feb 2026Award Received IBM Outstanding Technical Achievement Award for NL2Insights Impacting Products and Clients.
  • Feb 2026Milestone Multilingual Text2SQL capabilities are now available across all IBM Cloud and AWS production regions for IBM watsonx.data intelligence SaaS, supporting English and Japanese with more languages coming.
  • Jan 2026Patent Granted Site-wide optimization for mixed regression models and mixed control variables. Patent US12518174B2.
    D. T. Phan, N. H. Pham, L. M. Nguyen.
  • Jan 2026Workshop Black-Box Uncertainty Quantification for Large Language Models via Ensemble-of-Ensembles accepted for AAAI 2026 Workshop on Assessing and Improving Reliability of Foundation Models in the Real World (AAAI 2026 Workshop).
    W. Ma, D. Bhattacharjya, J. Lee, N. H. Pham, H. Kokel, Q. Ji.
  • Dec 2025Award Received IBM Growth Award for advancing Text2SQL service within watsonx.data intelligence.
  • Dec 2025Award IBM Research Accomplishments 2025 - A-level: NL2Insights achieved Product and Client-0 Adoption & Impact recognition. Our fully automated Text2SQL pipeline now powers watsonx.data, BI Assistant, and Process Mining, generating over 200,000 SQL queries across 1,000+ databases at enterprise scale.
  • Aug 2025Patent Application Generating structured query language queries from natural language inputs with schema enrichment
    M. Eyceoz, G. Rossiello, A. M. Gliozzo, M. R. Glass, N. Mihindukulasooriya, N. H. Pham, L. H. Vu, D. Subramanian, F. M. Chowdhury.
  • May 2025Award Received IBM Outstanding Technical Achievement Award for achieving first place on the BIRD Leaderboard with IBM Granite Text-to-SQL models.
  • May 2025Conference The Consistency Hypothesis in Uncertainty Quantification for Large Language Models accepted for Forty-First Conference on Uncertainty in Artificial Intelligence (UAI 2025).
    Q. Xiao, D. Bhattacharjya, B. Ganesan, R. Marinescu, K. Mirylenka, N. H. Pham, M. Glass, and J. Lee.
  • Mar 2025Patent Application Database Querying Using Natural Language Processing
    T. R. Dinger, A. M. Gliozzo, N. H. Pham, O. Hassanzadeh, D. Subramanian, L. Amini, G. Rossiello, M. F. M. Chowdhury, L. Vu, T. Kaple, M. Glass.
  • Dec 2024Award IBM Research Accomplishments 2024 - A-level: IBM Granite fine-tuned Text-to-SQL models sweep top spots in BIRD Leaderboard. Our Granite-20B and Granite-34B models achieved first place in both tracks, outperforming larger models like GPT-4 and GPT-4o. This success drove renewed interest from major industry players including Google, Alibaba, and ByteDance. Models were featured at THINK'24 and deployed on BAM and Watsonx.ai platforms.
  • Aug 2024Patent Application Fine-tuned generative model
    E. Lobo, N. H. Pham, L. Vu, T. Mummert, and D. Subramanian.
  • Jul 2024Milestone IBM’s text-to-SQL generator takes top place on a benchmark for handling complex database queries.
  • Apr 2024Career Update Promoted to Staff Research Scientist at IBM Thomas J. Watson Research Center.
  • Jan 2024Patent Application Combining data selection and reward functions for tuning large language models using reinforcement learning
    L. Vu, N. H. Pham, D. Subramanian, T. Mummert.
  • Sep 2023Conference Evaluating Robustness of Cooperative MARL: A Model-based Approach accepted for 2023 IEEE International Conference on Data Mining (ICDM).
    N. H. Pham, L. M. Nguyen, J. Chen, H. T. Lam, S. Das, T. W. Weng.
  • Aug 2023Award Received IBM 2022 Research Pat Goldberg Memorial Best Paper: A Hybrid Stochastic Optimization Framework for Stochastic Composite Nonconvex Optimization.
    Q. Tran-Dinh, N. H. Pham, D. T. Phan, and L. M. Nguyen.
  • Jun 2023Patent Application Generative modeling and representational learning from multi-sequence alignment and phylogenetic tree data
    T. L. Hoang, M. M. Galindo, G. Picco, M. Zayats, N. H. Pham, L. M. Nguyen, M. L. Sbodio, D. T. Phan, and V. L. Garcia.
  • Jun 2023Patent Application A novel meta-hyperparameter tuning system for RL using sequence model
    E. Lobo, N. H. Pham, D. Subramanian, and T. Pedapati.
  • Jun 2023Patent Application Reinforcement machine learning with multi-level agent search and hyperparameter optimization
    L. Vu, P. Kirchner, R. Marinescu, D. Subramanian, and N. H. Pham.
  • Feb 2023Tutorial Co-organizing a tutorial/lab forum LSHA2: Automated AI for Decision Optimization with Reinforcement Learning at AAAI 2023.
  • Feb 2023Other Session Chair of ML: Optimization 1 at AAAI 2023 on Feb 10, 2023.
  • Sep 2022Patent Application Adversarial Attacks for Improving Cooperative Multi-Agent Reinforcement Learning Systems
    N. H. Pham, L. M. Nguyen, J. Chen, T. L. Hoang, S. Das.
  • Jan 2022Career Update Started new role as Research Scientist at IBM Thomas J. Watson Research Center, Yorktown Heights, NY.
  • Dec 2021Milestone Completed PhD in Operations Research in the Department of Statistics and Operations Research at University of North Carolina at Chapel Hill. Dissertation
  • Sep 2021Conference FedDR – Randomized Douglas-Rachford Splitting Algorithms for Nonconvex Federated Composite Optimization accepted for the 35th Conference on Neural Information Processing Systems (NeurIPS 2021).
    Q. Tran-Dinh, N. H. Pham, D. T. Phan, and L. M. Nguyen.
  • Aug 2021Milestone Finished summer internship at Blue River Technology. Mentored by Ben Cline, supervised by Chris Padwick.
  • Jun 2021Preprint FedDR – Randomized Douglas-Rachford Splitting Algorithms for Nonconvex Federated Composite Optimization.
    Q. Tran-Dinh, N. H. Pham, D. T. Phan, and L. M. Nguyen.
  • May 2021Patent Application Site-Wide Optimization for Mixed Regression Models and Mixed Control Variables
    D. T. Phan, N. H. Pham, L. M. Nguyen.
  • May 2021Career Update Joined Blue River Technology as Machine Learning Intern.
  • Mar 2021Preprint Federated Learning with Randomized Douglas-Rachford Splitting Methods.
    Q. Tran-Dinh, N. H. Pham, D. T. Phan, and L. M. Nguyen.
  • Jan 2021Conference Regression Optimization for System-level Production Control accepted for 2021 American Control Conference (ACC).
    D. T. Phan, L. M. Nguyen, P. Murali, N. H. Pham, H. Liu, J. Kalagnanam.
  • Oct 2020Journal A Hybrid Stochastic Optimization Framework for Stochastic Composite Nonconvex Optimization accepted for Mathematical Programming.
    Q. Tran-Dinh, N. H. Pham, D. T. Phan, and L. M. Nguyen.
  • Aug 2020Milestone Finished summer internship at IBM Research. Mentored by Dzung T. Phan and Lam M. Nguyen, and supervised by Roman Vaculin.
  • Jun 2020Conference Stochastic Gauss-Newton Algorithms for Nonconvex Compositional Optimization accepted for the 37th International Conference on Machine Learning (ICML 2020).
    Q. Tran-Dinh, N. H. Pham, and L. M. Nguyen.
  • May 2020Career Update Joined IBM Research as Research Intern.
  • May 2020Journal ProxSARAH: An Efficient Algorithmic Framework For Stochastic Composite Nonconvex Optimization accepted for Journal of Machine Learning Research (JMLR).
    N. H. Pham, L. M. Nguyen, D. T. Phan, and Q. Tran-Dinh.
  • Mar 2020Preprint Finite-time analysis of stochastic gradient descent under markov randomness.
    T. T. Doan, L. M. Nguyen, N. H. Pham, and J. Romberg.
  • Feb 2020Preprint Stochastic Gauss-Newton Algorithms for Nonconvex Compositional Optimization.
    Q. Tran-Dinh, N. H. Pham, and L. M. Nguyen.
  • Feb 2020Preprint Convergence Rates of Accelerated Markov Gradient Descent with Applications in Reinforcement Learning.
    T. T. Doan, L. M. Nguyen, N. H. Pham, and J. Romberg.
  • Jan 2020Journal ProxSARAH: An Efficient Algorithmic Framework For Stochastic Composite Nonconvex Optimization accepted for Journal of Machine Learning Research (JMLR).
    N. H. Pham, L. M. Nguyen, D. T. Phan, and Q. Tran-Dinh.
  • Jan 2020Conference A Hybrid Stochastic Policy Gradient Algorithm for Reinforcement Learning accepted for the 23rd International Conference on Artificial Intelligence and Statistics (AISTATS 2020).
    N. H. Pham, L. M. Nguyen, D. T. Phan, P. H. Nguyen, M. van Dijk, and Q. Tran-Dinh.
  • Oct 2019Talk ProxSARAH: An Efficient Algorithmic Framework For Stochastic Composite Nonconvex Optimization. Oral Presentation: 2019 INFORMS Annual Meeting.
  • Jul 2019Preprint A Hybrid Stochastic Optimization Framework for Composite Nonconvex Optimization.
    Q. Tran-Dinh, N. H. Pham, D. T. Phan, and L. M. Nguyen.
  • Feb 2019Preprint ProxSARAH: An Efficient Algorithmic Framework For Stochastic Composite Nonconvex Optimization.
    N. H. Pham, L. M. Nguyen, D. T. Phan, and Q. Tran-Dinh.
  • Aug 2017Career Update Started PhD in Operations Research at University of North Carolina at Chapel Hill · Chapel Hill, NC, USA. Under supervision of professor Quoc Tran-Dinh.