Top-K Gated Macro-Actions for Neural Heuristic Search
Vedant Khandelwal et al. (2026). "Top-K Gated Macro-Actions for Neural Heuristic Search." Under review at SOCS 2026.
CV
Download PDFEducation
- Ph.D. in Computer Science, University of South Carolina, 2021 - 2026 (expected)
- Advisor: Amit Sheth
- Award: Best Poster Award at DISCOVER USC 2024, “Foundation Model For Pathfinding Problems” (>500 participants)
- M.S. in Computer Science, University of South Carolina, 2025
- Completed as part of direct Ph.D. program
- B.Tech. in Computer Science & Engineering, International Institute of Information Technology, Naya Raipur, 2020
- Award: Institute Gold Medal for Best All-Round Performance
Work Experience
- Summer Research Intern, IBM Thomas J. Watson Research Center, May 2025 - August 2025
- Developed a multi-agent fast–slow architecture (SOFAI-LM) with a metacognitive feedback loop to enhance LLM reasoning
- Demonstrated that the architecture helps an LLM outperform a standalone LRM in both accuracy and speed
- Identified that LLM history aids the LRM on local tasks (debugging) but hinders it on global tasks (graph coloring)
- Contributed to two patent applications and the accepted AAAI’26 Lab proposal
- Research Affiliate, Artificial Intelligence Institute, University of South Carolina, December 2019 - July 2020
- Developed a knowledge-infused machine learning pipeline for assessing the psychological impacts of pandemics using tweets
- Analyzed 12 billion COVID-19 tweets to study the spatiotemporal effect of the pandemic and policy decisions
Publications
Toward Neurosymbolic Reinforcement Learning via Editable Specifications
Vedant Khandelwal, Hong Yung Yip, and Amit Sheth (2026). "Toward Neurosymbolic Reinforcement Learning via Editable Specifications." AAAI-MAKE 2026.
Language Models Coupled with Metacognition Can Outperform Reasoning Models
Vedant Khandelwal, Francesca Rossi, Keerthiram Murugesan, Erik Miehling, Murray Campbell, Karthikeyan Natesan Ramamurthy, and Lior Horesh. (2025). "Language Models Coupled with Metacognition Can Outperform Reasoning Models." arXiv preprint arXiv:2508.17959 (under review at ICML 2026).
Inductive Logic Programming for Heuristic Search
Rojina Panta*, Vedant Khandelwal*, Celeste Veronese, Amit Sheth, Daniele Meli, and Forest Agostinelli. (2025). "Inductive Logic Programming for Heuristic Search." Accepted at both IJCLR 2025 (5th International Joint Conference on Learning & Reasoning) and the ICAPS 2025 Workshop on Planning and Reinforcement Learning (PRL). (* Equal contribution)
PDDLFuse: A Tool for Generating Diverse Planning Domains
Vedant Khandelwal, Amit Sheth, and Forest Agostinelli. (2025). "PDDLFuse: A Tool for Generating Diverse Planning Domains." GenPlan Workshop, AAAI 2025.
Towards Learning Foundation Models for Heuristic Functions to Solve Pathfinding Problems
Vedant Khandelwal, Amit Sheth, and Forest Agostinelli. (2025). "Towards Learning Foundation Models for Heuristic Functions to Solve Pathfinding Problems." GenPlan Workshop, AAAI 2025.
NeuroSymbolic Knowledge-Grounded Planning and Reasoning in Artificial Intelligence Systems
Amit Sheth, Vedant Khandelwal, Kaushik Roy, Vishal Pallagani, and Megha Chakraborty. (2025). "NeuroSymbolic Knowledge-Grounded Planning and Reasoning in Artificial Intelligence Systems." IEEE Intelligent Systems, pp. 27-34.
AI-Augmented Search for Systematic Reviews: A Comparative Analysis
Valerie Vera*, Vedant Khandelwal*, Kaushik Roy, Ritvik Garimella, Harshul Surana, and Amit Sheth. (2025). "AI-Augmented Search for Systematic Reviews: A Comparative Analysis." Proceedings of the Association for Information Science and Technology 62, no. 1: 705-717. (* Equal contribution)
A Neurosymbolic Fast and Slow Architecture for Graph Coloring
Vedant Khandelwal, Vishal Pallagani, Biplav Srivastava, and Francesca Rossi. (2025). "A Neurosymbolic Fast and Slow Architecture for Graph Coloring." Twelfth Annual Conference on Advances in Cognitive Systems (ACS).
A Domain-Agnostic Neurosymbolic Approach for Big Social Data Analysis: Evaluating Mental Health Sentiment on Social Media during COVID-19
Vedant Khandelwal, Manas Gaur, Ugur Kursuncu, Valerie Shalin, and Amit Sheth. (2024). "A Domain-Agnostic Neurosymbolic Approach for Big Social Data Analysis: Evaluating Mental Health Sentiment on Social Media during COVID-19." Proceedings of the IEEE International Conference on Big Data.
Specifying goals to deep neural networks with answer set programming
Forest Agostinelli, Rojina Panta, and Vedant Khandelwal. (2024). "Specifying goals to deep neural networks with answer set programming." Proceedings of the ICAPS, vol. 34, pp. 2-10.
GEAR-Up: Generative AI and External Knowledge-based Retrieval Upgrading Scholarly Article Searches for Systematic Reviews
Kaushik Roy, Vedant Khandelwal, Harshul Surana, Valerie Vera, Amit Sheth, and Heather Heckman. (2024). "GEAR-Up: Generative AI and External Knowledge-based Retrieval Upgrading Scholarly Article Searches for Systematic Reviews." Proceedings of the AAAI Conference on Artificial Intelligence.
Demo Alleviate: Demonstrating Artificial Intelligence Enabled Virtual Assistance for Telehealth: The Mental Health Case
Kaushik Roy, Vedant Khandelwal, Raxit Goswami, Nathan Dolbir, Jinendra Malekar, and Amit Sheth. (2023). "Demo Alleviate: Demonstrating Artificial Intelligence Enabled Virtual Assistance for Telehealth: The Mental Health Case." Proceedings of the AAAI Conference on Artificial Intelligence 37(13): 16289-16290.
ALLURE: A Multi-Modal Guided Environment for Helping Children Learn to Solve a Rubik’s Cube with Automatic Solving and Interactive Explanations
Kausik Lakkaraju, Thahimum Hassan, Vedant Khandelwal, Prathamjeet Singh, Cassidy Bradley, Ronak Shah, Forest Agostinelli, Biplav Srivastava, and Dezhi Wu. (2022). "ALLURE: A Multi-Modal Guided Environment for Helping Children Learn to Solve a Rubik's Cube with Automatic Solving and Interactive Explanations." Proceedings of the AAAI Conference on Artificial Intelligence, vol. 36, no. 11, pp. 13185-13187.
Explainable Pathfinding for Inscrutable Planners with Inductive Logic Programming
Forest Agostinelli, Rojina Panta, Vedant Khandelwal, Biplav Srivastava, Bharath Chandra Muppasani, Kausik Lakkaraju, and Dezhi Wu. (2022). "Explainable Pathfinding for Inscrutable Planners with Inductive Logic Programming." ICAPS 2022 Workshop on Explainable AI Planning.
“Is depression related to cannabis?”: A knowledge-infused model for entity and relation extraction with limited Supervision
Kaushik Roy, Usha Lokala, Vedant Khandelwal, and Amit Sheth. (2021). "Is depression related to cannabis?: A knowledge-infused model for entity and relation extraction with limited Supervision." Proceedings of the AAAI 2021 Spring Symposium on Combining Machine Learning and Knowledge Engineering (AAAI-MAKE 2021).
Tutorials, Labs & Invited Talks
SOFAI-LM: A Cognitive Architecture for Building Efficient and Reliable Reasoning Systems with LLMs
Quarter Day Lab at AAAI 2026
Vedant Khandelwal, Francesca Rossi, Vishal Pallagani, Keerthiram Murugesan, and Lior Horesh
Website | Slides | Code | pdfTowards Learning Foundation Models for Heuristic Functions to Solve Pathfinding Problems
Contributing Talk at the GenPlan Workshop, AAAI 2025, February 2025
Vedant Khandelwal, Amit Sheth, and Forest Agostinelli
SlidesSOFAI Lab: A Hands-On Guide to Building Neurosymbolic Systems with Metacognitive Control
Contributing Presenter, AAAI 2025 Quarter Day Labs, February 2025
Vedant Khandelwal
SlidesNeuro-Symbolic AI for Deep Analysis of Social Media Big Data (Tutorial)
IEEE International Conference on Big Data, December 2024
Vedant Khandelwal, Manas Gaur, Ugur Kursuncu, Valerie Shalin, and Amit Sheth
Slides | Website | pdfIntroduction to Transformers: Theory and Practical Applications
Guest Lecture at CSCE 180 - Artificial Intelligence for All, University of South Carolina, September 2024
Vedant Khandelwal
SlidesFoundation Models
Guest Lecture at CSCE 580 - Artificial Intelligence, University of South Carolina, April 2024
Vedant Khandelwal
Slides
Panels & Seminars
A Neurosymbolic Fast and Slow Architecture for Graph Coloring
Presented at the CP/ML Bridge during AAAI 2025, February 2025
Vedant Khandelwal
SlidesA Brief Primer on AI-Based Tools and Generative AI Applications
Panelist at the seminar “From Chatbots to Generative AI: How to Ethically and Efficiently Advise Clients on the Use of AI-Based Technology in Everyday Practice”
2025 South Carolina Bar Convention, January 16-19, 2025
Vedant Khandelwal
SlidesPsychidemic: Measuring the Spatio-Temporal Psychological Impact of Novel Coronavirus with a Social Quality Index
CSCE 791 - Seminar in Advances in Computing, University of South Carolina, April 2021
Vedant Khandelwal
Slides
Patents
U.S. Patent 12,332,873 (Issued June 17, 2025)
“Multimodal retrieval and execution monitoring using rich recipe representation”
Biplav Srivastava, Vishal Pallagani, Revathy Venkataramanan, Vedant Khandelwal, and Kausik LakkarajuU.S. Patent 12,067,983 (Issued August 20, 2024)
“Robust useful and general task-oriented virtual assistants”
Biplav Srivastava, Kausik Lakkaraju, Revathy Venkataramanan, Vishal Pallagani, Vedant Khandelwal, and Hong Yung Yip