About Me
I am a PhD student in Electrical Engineering at Columbia University, working with Prof. Tanvir Ahmed Khan. My research focuses on profile guided optimization, optimizing power management systems, and data provenance for database systems. I am particularly interested in optimizing the performance and power efficiency of GPUs and other accelerators, both from a compiler perspective and power management. My ultimate goal is to create a generalized system capable of optimizing any combination of heterogeneous systems for any application.
Education
-
Columbia University
PhD in Electrical Engineering
New York, USA | Sept. 2024 - Present
-
Bangladesh University of Engineering and Technology
Masters of Science in Computer Science and Engineering
Palashi, Dhaka, Bangladesh | July 2021 - Aug. 2024
-
Bangladesh University of Engineering and Technology
Bachelor of Science in Computer Science and Engineering
Palashi, Dhaka, Bangladesh | Mar. 2016 - Feb. 2021
Class Rank: 1st
Experience
-
Graduate Research and Teaching Assistant
Columbia University, New York, USA
Sept. 2024 - Present
-
Lecturer, Department of CSE
Bangladesh University of Engineering and Technology, Dhaka, Bangladesh
Sept. 2021 - Aug. 2024
Courses Instructed: Operating System, Database Management System, Software Development, Computer Graphics, Computer Programming, Digital Logic Design
-
Lecturer, Department of CSE
Brac University, Mohakhali, Dhaka, Bangladesh
Jun. 2021 - Sept. 2021
Courses Instructed: Discrete Mathematics, Algorithms, Computer Graphics, Microprocessors
-
Lecturer, Department of CSE
United International University, Badda, Dhaka, Bangladesh
Feb. 2021 - Jun. 2021
Courses Instructed: Database Management System, Computer Graphics
Publications
-
Reverse Engineering DVFS Mechanisms
2025 IEEE International Symposium on Hardware Oriented Security and Trust (HOST), San Jose, CA, USA, 2025, https://doi.ieeecomputersociety.org/10.1109/HOST64725.2025.11050037
-
Computing the Largest Common Almost-Increasing Subsequence
Theoretical Computer Science, Volume 930, 2022, Pages 157-178, ISSN 0304-3975, https://doi.org/10.1016/j.tcs.2022.07.021
-
The Disproportionate Impact of COVID-19 among Undocumented Immigrants and Racial Minorities in the US
International Journal of Environmental Research and Public Health, Volume 18(23): 12708, 2021, PubMedID 34886437, ISSN 1660-4601, https://doi.org/10.3390/ijerph182312708
Projects
-
Optimizing GPU energy efficiency for LLM
Power Management | Dec. 2024 - Present
Characterizing the hardware involved to find the breakdown of energy consumption in different components and proposing micro-architectural and/or software based mechanisms to manage different hardware components optimally
-
Profile Inference
Compiler | Sept. 2024 - Present
Inferring stale profile for better profile guided optimization of the binary
Work under submission
-
Scalable Power Management
Power Management | Dec. 2024 - Present
Devising power management protocol for large scale interconnected system components with very low latency
Work under submission
-
Data Provenance
Database, Compiler | Jan. 2025 - Present
Data Provenance using compiler-based instrumentation
-
Parallelization Suggestion for Python Programs
Python, Systems | Jan. 2022 - Aug. 2024
Profiling Python programs to provide suggestions to run it in a distributed environment. Prepared a prototype and testing it. The prototype performs the parallelization task as fast as a manually parallelized program for a set of inputs.
-
Incorporating local search into ACO based many-objective optimization
Meta-heuristics | Apr. 2022 - Aug. 2022
Incorporated local search into one type of ACO-based many-objective optimization algorithms, \(iMOACO_{\mathbb{R}}\). Tested the new algorithm against \(iMOACO_{\mathbb{R}}\) and found out it sometimes gives us two times improvement and always stays competitive.
-
A Hybrid Approach to DNS Cache Poisoning Security
Networking Protocols | Dec. 2021 - Mar. 2022
Proposed a novel hybrid approach to DNS cache poisoning. In contrast to classical DNS poisoning attacks that target resolvers, our proposed attack targets client-side caches with the help of an unprivileged malware residing in the client machine. To improve the reliability of our attack, we also exploit the Response Rate Limiting (RRL) feature present in most nameservers. Combining these two ideas gives us an attack method that is surprisingly effective.
Scholarships, Awards, and Grants
- 2022: Crest of Honor, Highest CGPA in the department, presented by BUET alumni association
- 2016-2021: University Merit Scholarship, Bangladesh University of Engineering and Technology
- 2016-2021: Dean's List Scholarship, Bangladesh University of Engineering and Technology
Service
- Organizing Committee Member of NSysS 2021, 2022, 2023