Hello, this is Amit.
I am a graduate student at Purdue Computer Science, investigating the capabilities of LLM in solving graph problems. My research focus is primarily deep learning and I have experience in developing deep learning architectures that approach real-life problems specifically involving Graph Neural Networks (GNN). I worked with Prof. Pan Li for a while on the Graph Anomaly Detection problem, funded by Sony, Belgium.
Apart from my research and academic activities, I enjoy following and analyzing cricket, watching movies and TV series, and playing badminton and card games.
I am available for applied science research internships for Summer 2025 involving LLM and GNN (May 2025-August 2025).
Past Experiences and Education ...
Before joining Purdue CS in the Fall of 2022, I worked as a research assistant in the Artificial Intelligence and Cybernetics Lab (AGenCy Lab) at the Department of Computer Science and Engineering, Independent University Bangladesh where I worked on several research projects related to deep learning on graph data supported by ICT Division, Government of Bangladesh. Thanks to my supervisors Dr. Amin Ahsan Ali and Dr. AKM Mahbubur Rahman. Also, I have experience working as a software engineer at Tiger IT Bangladesh Limited where I worked as a backend developer on applied computer vision projects from real-time video data.
I received my Bachelor of Science (B.Sc) degree in Computer Science and Engineering from the University of Dhaka, Dhaka, Bangladesh in August 2020 and secured the 1st position in the class of 2019. In my undergraduate days, I was involved in competitive programming and solved 1000+ problems in different online judges [stopstalk profile]. My undergraduate research topic was Utility-based Graph Mining, supervised by Dr. Chowdhury Farhan Ahmed and Md. Ashraful Islam .
📰 News
- [Oct 2023] GAD-EBM got accepted in the NeurIPS 2023 GLFRONTIERS Workshop .
- [Oct 2023] GAD-NR got accepted in the WSDM-2024 .
- [Aug 2022] Started my grad school at Purdue Computer Science.
- [Apr 2022] Received Ross Fellowship and selected for Research Assistantship (RA) position by the Purdue University - Computer Science.
📝 Publications
-
GAD-NR: Graph Anomaly Detection via Neighborhood Reconstruction
Amit Roy, Juan Shu, Jia Li, Carl Yang, Olivier Elshocht, Jeroen Smeets, Pan Li
The 17th ACM International Conference on Web Search and Data Mining (WSDM-2024)
March 4th-8th, 2024, Mérida, Yucatán, Mexico
[ Proceedings ] [ Paper ] [ Poster ] [ Presentation ] [ Code ] [ Video ] [ PyGod ] [ BibTex ]
Before joining Purdue CS:
-
Unified Spatio-Temporal Modeling for Traffic Forecasting using Graph Neural Network
Amit Roy*, Kashob Kumar Roy*, Amin Ahsan Ali, M Ashraful Amin and A K M Mahbubur Rahman
The International Joint Conference on Neural Networks (IJCNN-2021), Publisher: IEEE,
Virtual Event, July 18-22, 2021
[ Proceedings ] [ Paper ] [ Code ] [ Video ] [ BibTex ] -
SST-GNN: Simplified Spatio-temporal Traffic Forecasting Model using Graph Neural Network
Amit Roy*, Kashob Kumar Roy*, Amin Ahsan Ali, M Ashraful Amin and A K M Mahbubur Rahman,
The 25th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD-2021),
Publisher: Springer, May 11-14, 2021, Delhi, India (Virtual).
[ Proceedings ] [ Paper ] [ Code ] [ Video ] [ BibTex ] -
Node Embedding using Mutual Information and Self-Supervision based Bi-level Aggregation
Kashob Kumar Roy*, Amit Roy*, Amin Ahsan Ali, M Ashraful Amin and A K M Mahbubur Rahman,
The International Joint Conference on Neural Networks (IJCNN-2021), Publisher: IEEE,
Virtual Event, July 18-22, 2021
[ Proceedings ] [ Paper ] [ Code ] [ Video ] [ BibTex ] -
Structure-Aware Hierarchical Graph Pooling using Information Bottleneck
Kashob Kumar Roy*, Amit Roy*, Amin Ahsan Ali, M Ashraful Amin and A K M Mahbubur Rahman
The International Joint Conference on Neural Networks (IJCNN-2021), Publisher: IEEE,
Virtual Event, July 18-22, 2021
[ Proceedings ] [ Paper ] [ Code ] [ Video ] [ BibTex ] -
UGMINE: Utility-based Graph Mining
Md. Tanvir Alam, Amit Roy, Chowdhury Farhan Ahmed, Md. Ashraful Islam, Carson K. Leung,
Applied Intelligence Journal, Impact Factor: 5.086 , Publisher: Springer, April 12, 2022
[ Proceedings ] [ Paper ] [Code] [ BibTex ] -
Mining High Utility Subgraphs
Md. Tanvir Alam, Amit Roy, Chowdhury Farhan Ahmed, Md. Ashraful Islam, Carson K. Leung,
UDML 2021: 4th Workshop on Utility Driven Mining and Learning @ICDM 2021, Publisher: IEEE
December 7, 2021, Auckland, New Zealand (Virtual)
[ Proceedings ] [ Paper ] [Code] [ BibTex ]
* indicates Equal Contribution
🎓 Education
-
Graduate Student in Department of Computer Science
Purdue University - West Lafayette, IN, USA
August, 2022 - Present
(MS Requirements fulfilled on May 2024)
Grad School Webpage -
B.Sc in Computer Science and Engineering
January, 2016 - January, 2020
CGPA: 3.96 out of 4.00 (Class Rank: 1st)
University of Dhaka, Dhaka, Bangladesh
Undergraduate Curriculum
💻 Experience
Teaching Experience
Teaching Assistant, Department of Computer Science, Purdue University
-
CS37300: Data Mining and Machine Learning, Fall 2024
Instructor: Prof. Dan Goldwasser, Prof. Bruno Ribeiro -
CS37300: Data Mining and Machine Learning, Spring 2024
Instructor: Prof. Steve Hanneke and Prof. Ruqi Zhang -
CS57800: Statistical Machine Learning, Fall 2023
Instructor: Prof. Ruqi Zhang
Industry Experience
-
Software Engineer [December 2020 - August 2022]
Tiger IT Bangladesh Limited
Dhaka, Bangladesh
Research Experience
-
Research Assistant [Feb 2020 - June 2022]
Artificial Intelligence and Cybernetics Lab,
Independent University, Bangladesh (AGenCy Lab @ IUB)
Dhaka, Bangladesh
🎖️ Honors and Awards
-
Professor Dr. M. Lutfar Rahman Award
Department of CSE, University of Dhaka
For securing the highest CGPA in the Bachelor of Science Final Examination of 2019 [ Credentials ]
-
University Grants Commission, Bangladesh Scolarship
For holding the first position in the Bachelor of Science Final Examination of 2019 in the
Faculty of Engineering and Technology, University of Dhaka. [ Credentials ]