Sunandita Patra
I am a Research Lead at J. P. Morgan AI Research. I have worked as a Postdoc and Ph.D. student in the Department of Computer Science, University of Maryland, College Park, where I defended my PhD in Summer 2020. My PhD advisor was Prof. Dana Nau. Before this, I graduated with Bachelor's and Master's of Technology degrees in Computer Science and Engineering at Indian Institute of Technology, Kharagpur, and worked as a Software Engineer at Microsoft.
Areas of Interest:
acting, planning, and machine learning;
applications of AI in finance, robotics, and cybersecurity;
search and optimization;
Research Interests
My research interest is the integration of acting, planning and machine learning, to build deliberative actors. My work has focused on robotics, cybersecurity, and finance environments. For an actor to operate in such a real-world environment, a critical issue is to account for unpredictable and evolving conditions with dynamic events, dead ends, concurrent tasks, sensing information from the environment, nondeterministic outcomes of actions, and uncertainty in different forms. However, most conventional AI systems do not incorporate such realistic scenarios, which requires developing novel integrated algorithms that incorporate within a single framework: (a) deliberative acting, (b) online planning, and (c) learning from the actor's experiences. In my published work, I have leveraged structured representations, such as hierarchical task networks, temporal goal networks, and operational models, to aid the actor's decision-making process. This enables unique capabilities of effective coordination between the planner, the acting engine, and the learning component to achieve a common goal. Given the complementary strengths of structured reasoning in planning algorithms and the flexible, data-driven capabilities of large language models and deep learning, my goal is to develop hybrid approaches that harness the strength of both paradigms.
My interest also lies in hierarchical planning, goal reasoning and recognition, scheduling under uncertainty, monte-carlo tree search, and planning in cooperative multi-agent systems. I want to synthesize planning and acting systems that can behave intelligently across a wide range of problem domains with imperfect sensors and incomplete knowledge of the environment. Toward this goal, I have adapted the developed integrated planning and learning algorithms for real-world scenarios, such as robotics, security, and finance applications.
Research Projects
Developed acting and online planning algorithms for dynamic environments in which both planning (coming up with a strategy to accomplish a task) and acting (carrying out actions in the real world) use the same hierarchical operational models. This has several benefits with respect to consistency verification of the different models and closed-loop online decision-making. Developed simulated test domains to evaluate the performance of the algorithms in terms of three newly designed metrics: efficiency, retry ratio, and success ratio to evaluate acting and planning systems.
Selected Publications (For full list, please see this link):
(6) Sunandita Patra, M Ghallab, D Nau, P Traverso. Acting and Planning Using Operational Models. AAAI. 2019.
(5) R Li, Sunandita Patra, DS Nau. Decentralized Refinement Planning and Acting. ICAPS. 2021.
(4) Y Bansod, Sunandita Patra, D Nau, M Roberts. HTN Replanning from the Middle. The International FLAIRS Conference Proceedings 35. 2022.
(3) Sunandita Patra, P Traverso, M Ghallab, D Nau. Coordination and Control of Hierarchically Organized Interacting Agents. 34th Florida Artificial Intelligence Research Society Conference (FLAIRS-34). 2021.
(2) D Nau, Y Bansod, Sunandita Patra, M Roberts, R Li. GTPyhop: A hierarchical goal+ task planner implemented in Python. ICAPS Workshop on Hierarchical Planning (HPlan). 2021.
(1) Sunandita Patra, Malik Ghallab, Dana Nau, Paolo Traverso. APE: An Acting and Planning Engine. Journal of Advances in Cognitive Systems. Volume 7. 2018.
Integration of acting and planning (2015 - current)
At University of Maryland, College Park in collaboration with LAAS-CNRS, France, FBK, Italy and Naval Research Labs, USA
Integration of acting, planning and machine learning (2019 - current)
At University of Maryland, College Park in collaboration with LAAS-CNRS, France, FBK, Italy and Naval Research Labs, USA
Integrated machine learning with online acting and planning to learn: (a) the optimal operational model to accomplish a task; (b) a domain-independent heuristic for our hierarchical operational models, using multi-layered perceptrons.
Developed a goal-biased curriculum for temporal goal networks in dynamic environments in which hierarchical planning and reinforcement learning are combined online, using a hierarchical goal network formalism.
Selected Publications (For full list, please see this link):
(4) Sunandita Patra, J Mason, M Ghallab, D Nau, P Traverso. Deliberative acting, planning and learning with hierarchical operational models. Artificial Intelligence Journal (AIJ), Volume 299. 2021.
(3) Sunandita Patra, J Mason, A Kumar, M Ghallab, P Traverso, D Nau. Integrating Acting, Planning and Learning in Hierarchical Operational Models. ICAPS. 2020. Best Student Paper Honorable Mention Award at ICAPS.
(2) Sunandita Patra, M Cavolowsky, O Kulaksizoglu, R Li, L Hiatt, M Roberts, D Nau. A Hierarchical Goal-Biased Curriculum for Training Reinforcement Learning. The International FLAIRS Conference Proceedings 35. 2022.
(1) Sunandita Patra. Acting, Planning, and Learning Using Hierarchical Operational Models. PhD Thesis at University of Maryland, College Park. 2020.
AI in finance (2021 - current)
At JPMorgan AI Research
Working in the integration of planning, acting and learning, and applications of AI in finance. Developed a framework to solve financial trading problems (asset allocation and trade execution) using AI planning and reinforcement learning algorithm. For more details, please see the video below:
Selected Publications (For full list, please see this link):
(2) Sunandita Patra, M Mahfouz, S Gopalakrishnan, D Magazzeni, M Veloso. FinRDDL: Can AIPlanning be used for Quantitative Finance Problems? ICAPS Financial Planning Workshop (FinPlan).2023.
(1) M Mahfouz, S Gopalakrishnan, M Suau, Sunandita Patra, D Mandic, D Magazzeni, M Veloso.Towards Asset Allocation Using Behavioural Cloning and Reinforcement Learning. AAAI Bridge: AI for Financial Services. 2023.
Acting and planning for network security (2019 - 2021)
At University of Maryland, College Park in collaboration with Naval Research Labs, USA
Worked in a research collaboration with the Naval Research Labs to use our refinement acting engine and online planner to defend software-defined networks against high-volume fast-paced incoming attacks.
Selected Publications (For full list, please see this link):
(2) A Velazquez, B Montrose, M Li, J Luo, M Kang, Sunandita Patra, D Nau. ACRS4SDN: An Autonomous Cyber Response System for Software-Defined Networks. Naval Research LaboratoryTechnical Report. 2022.
(1) Sunandita Patra, A Velazquez, M Kang, D Nau. Using online planning and acting to recover from cyberattacks on software-defined networks. AAAI. 2021.
Search and Optimization (2013 - current)
At JPMorgan AI Research, University of Maryland, College Park, Indian Institute of Technology, Kharagpur,
TU Munich, Germany
I have worked on several projects related to heuristic search and optimization. I mention three of them here:
(3) Developed a scheduling algorithm to optimize the conflicting objectives of cost and quality of service of jobs with uncertain duration and resource usage on JPMorgan’s grid compute cluster.
(2) Developed efficient anytime algorithms for multi-objective optimization which incrementally explores the state space with given contracts (intervals for reporting), works without restarting, and dynamically adapts for the next iteration.
(1) Designed and implemented algorithms for co-design of controller and scheduler parameters for embedded systems with multiple control loops and a hierarchical scheduler, by calculating optimal delays in each control application.
Selected Publications (For full list, please see this link):
(3) Sunandita Patra. Anytime Multi-Objective Optimization and Co-design of Controller and Scheduler for Optimal Control Performance. Master's Thesis at IIT Kharagpur. 2014.
(2) Sunandita Patra. Anytime Contract Heuristic Search Methods for Single and Bi-Objective Optimization Problems. Bachelors Thesis at IIT Kharagpur. 2013.
(1) Sunandita Patra, SG Vadlamudi, PP Chakrabarti. Anytime contract search. Research and Development in Intelligent Systems. Springer. 2013.
Selected Publications
(For full list, please see this link):
Sunandita Patra, Keshav Ramani, Daniel Borrajo, Sriram Gopalakrishnan. Generating Domain Specific Natural Language SAT Reasoning Datasets. Neurips Efficient Reasoning Workshop. 2025.
Oscar Lima*, Marc Vinci*, Sunandita Patra*, Sebastian Stock, Joachim Hertzberg, Martin Atzmueller, Malik Ghallab, Dana Nau, Paolo Traverso. Acting and Planning with Hierarchical Operational Models on a Mobile Robot: A Study with RAE+UPOM. European Conference on Mobile Robots (ECMR). 2025. *equal contribution. Best Student Paper Finalist (Top 3) Award.
Sunandita Patra, Mehtab Pathan, Mahmoud Mahfouz, Parisa Zehtabi, Wided Ouaja, Daniele Magazzeni, Manuela Veloso. Capacity Planning and Scheduling for Jobs with Uncertainty in Resource Usage and Duration. The Journal of Supercomputing 80 (15). 2024.
Sunandita Patra, Mahmoud Mahfouz, Sriram Gopalakrishnan, Danilo Mandic, Daniele Magazzeni, Manuela Veloso. FinRDDL: Can AI Planning be used for Financial Trading Problems? ICAPS Financial Planning (FinPlan) Workshop. 2023.
Sunandita Patra, Mark Cavolowsky, Onur Kulaksizoglu, Ruoxi Li, Laura Hiatt, Mark Roberts, Dana Nau. A Hierarchical Goal-Biased Curriculum for Training Reinforcement Learning. FLAIRS. 2022. [paper]
Sunandita Patra, James Mason, Malik Ghallab, Paolo Traverso, Dana Nau. Deliberative Acting, Online Planning and Learning with Hierarchical Operational Models. Artificial Intelligence (AIJ). 2021. [paper]
Ruoxi Li, Sunandita Patra, Dana Nau. Decentralized Refinement Planning and Acting. ICAPS 2021. [paper]
Sunandita Patra, Alex Velazquez, Myong Kang, Dana Nau. Using Online Planning and Acting to Recover from Cyberattacks on Software-defined Networks. IAAI 2021. [paper][poster]
Sunandita Patra, James Mason, Amit Kumar, Malik Ghallab, Dana Nau, Paolo Traverso. Integrating Acting, Planning, and Learning in Hierarchical Operational Models. ICAPS 2020. Best Student Paper Honorable Mention Award. [paper][poster]
Sunandita Patra, Malik Ghallab, Dana Nau, Paolo Traverso. Acting and Planning Using Operational Models. AAAI 2019. [paper][poster]
Sunandita Patra, Malik Ghallab, Dana Nau and Paolo Traverso. APE: An Acting and Planning Engine. Journal of Advances in Cognitive Systems, 2018. [paper]
Professional Activities
Invited Talks:
Nov 2022 DFKI, Germany
May 2021 JPMorgan AI Research, USA
Jun 2021 Xerox PARC, USA
May 2021 Jet Propulsion Laboratory (NASA JPL), USA [slides]
Dec 2020 Naval Warfare Center, USA [slides]
Oct 2020 Summer School at ICAPS Conference (International) [video]
Reviewer:
Artificial Intelligence Journal (AIJ) (2025)
AAAI conference (2025)
Neurips Conference (2025)
AIStats Conference (2024)
ICAPS Conference (2021 - 2024)
ECAI Conference (2023 - 2025)
FLAIRS Conference (2021 - 2024)
ICAPS Hierarchical Planning Workshop (2020 - 2023)
ICAPS Integrated Acting, Planning and Execution Workshop (2020 - 2023)
ICAPS Workshop on International Planning Competition (WIPC) (2021)
MACH (Machine Learning Journal, Springer) (2021)
Workshop on the Algorithmic Foundations of Robotics (WAFR) (2020)
Organizer:
ICAIF Tutorial on Evaluating and Rating AI Systems for Trust, and Its Applications to Finance (2024)
ICAPS Workshop on Integrated Acting, Planning and Execution (2021 - 2023)
Honors and Awards
2025 Received Best Student Paper Finalist (Top 3) Award at ECMR Conference
2020 Received Best Student Paper Honorable Mention Award at ICAPS Conference
2019 Received the AAAI student travel scholarship
2018 Got accepted for the French-American Doctoral Exchange Program (FADEx), France
2018 Received the Goldhaber Travel Grant at University of Maryland, USA
2016 Received University of Maryland Dean’s Fellowship, USA
2015 Received University of Maryland Dean’s Fellowship, USA
2013 DAAD-WISE (Working Internships in Science and Engineering) scholarship, Germany
2012 Goralal Syngal Scholarship (for being one the best CGPA holders), IIT Kharagpur
2011 Goralal Syngal Scholarship (for being one the best CGPA holders), IIT Kharagpur
2009 Cleared IIT-Joint Entrance Examination with an All India Rank of 1264, India
2009 Cleared All India Entrance Examination (written & interview) for admission to Indian Statistical Institute, India
2007 Cleared Regional Mathematical Olympiad (RMO), attended Indian National Mathematical Olympiad (INMO) training camp at ISI Kolkata, India
