Saurabh Singh Rajput is a Ph.D. student with the SMART-lab at Dalhousie University, under the supervision of Dr. Tushar Sharma. My research lies at the intersection of artificial intelligence, hardware processors and software systems. My goal? To develop innovative techniques that help profile, analyze and enhance the energy and compute efficiency of deep learning systems. By doing so, I aim to contribute to the creation of a greener AI, ensuring our technological advancements support a sustainable future. Before this, I was busy adding software to your finances at Fidelity Investments.
News
- [Dec. 2025] “FlipFlop: A Static Analysis-based Energy Optimization Framework for GPU Kernels”. Accepted in ACM/IEEE International Conference on Software Engineering (ICSE 2026) Research Track. Dec 2025.
- [Dec. 2025] “Tu(r)ning AI Green: Exploring Energy Efficiency Cascading with Orthogonal Optimizations”. Accepted in IEEE Software Mar 2026. Dec 2025.
- [Nov. 2025] “Full-spectrum Energy Profiling: Methods, Challenges, and Applications” International Conference on Collaborative Advances in Software and COmputiNg (CASCON) Tutorial. Nov 2025.
- [Oct 2025] “Why Attention Fails: A Taxonomy of Faults in Attention-Based Neural Networks”. Accepted in ACM/IEEE International Conference on Software Engineering (ICSE 2026) Research Track. Oct 2025.
- [May. 2025] Awarded Best Talk Award Consortium for Software Engineering Research (CSER). May 2025.
- [May. 2025] “Towards Fine-grained energy measurement of Software” Research Talk ACM/IEEE International Conference on Software Engineering (ICSE 2025), May 2025
- [May 2025] “Towards Fine-grained energy measurement of Software” Research Talk Consortium for Software Engineering Research (CSER), May 2025
- [December. 2024] “Enhancing Energy-Awareness in Deep Learning through Fine-Grained Energy Measurement”, Accepted by ACM/IEEE International Conference on Software Engineering (ICSE 2025) Journal First Track
- [November. 2024] Passed the Ph.D. Candidacy Exam. Wohooooo!
- [November. 2024] “COMET: Generating Commit Messages using Delta Graph Context Representation”, Accepted by Journal of Systems and Software (JSS)
- [November. 2024] Presented research poster on “Towards Energy-Efficient CUDA Kernels: A Predictive Modeling Approach” Accepted at 34th International Conference on Collaborative Advances in Software and COmputiNg (CASCON) Toronto, Ontario.
- [June. 2024] “Enhancing Energy-Awareness in Deep Learning through Fine-Grained Energy Measurement”, Accepted by ACM Transactions on Software Engineering and Methodology (TOSEM)
- [June. 2024] Presented research poster on “Investigating the Energy Efficiency of Popular Quantization Techniques for AI Models” Accepted at Software Engineering for Machine Learning Applications (SEMLA) international symposium, Montreal, Quebec.
- [May. 2024] Presented research poster on “FlipFlop: Predictive Power Modeling and Optimization for Energy-Efficient GPU Computing” Accepted at Dalhousie AI Symposium, Halifax, Nova Scotia.
- [April. 2024] Presented research poster on “FlipFlop: Predictive Power Modeling and Optimization for Energy-Efficient GPU Computing” Accepted at Smart Energy Conference, Halifax, Nova Scotia.
- [Mar. 2024] “Pursuit of Energy-efficient AI: Benchmarking Emerging Neural Network Quantization Methods”, Accepted at International Workshop on Green and Sustainable Software (GREENS’24) - Co-Located with ICSA24
- [Jan. 2024] “Greenlight: Highlighting TensorFlow APIs Energy Footprint”, Accepted in MSR (Data/Tools track) 2024.
- [Nov. 2023] Junior PC for MSR 2024 Technical Track.
Publications
- “FlipFlop: A Static Analysis-based Energy Optimization Framework for GPU Kernels”. Accepted in ACM/IEEE International Conference on Software Engineering (ICSE 2026) Research Track. Dec 2025. Preprint
- “Tu(r)ning AI Green: Exploring Energy Efficiency Cascading with Orthogonal Optimizations”. Accepted in IEEE Software Mar 2026. Dec 2025. Preprint
- “Why Attention Fails: A Taxonomy of Faults in Attention-Based Neural Networks”. Accepted in ACM/IEEE International Conference on Software Engineering (ICSE 2026) Research Track. Oct 2025. Preprint
- “Enhancing Energy-Awareness in Deep Learning through Fine-Grained Energy Measurement”, Accepted by ACM Transactions on Software Engineering and Methodology (TOSEM) Preprint
- “Pursuit of Energy-efficient AI: Benchmarking Emerging Neural Network Quantization Methods”, Accepted at International Workshop on Green and Sustainable Software (GREENS’24) - Co-Located with International Conference on Software Architecture (ICSA) Preprint
- “Greenlight: Highlighting TensorFlow APIs Energy Footprint”, Accepted in International Mining Software Repositories Conference MSR (Data/Tools track) 2024. Preprint
- “COMET: Generating Commit Messages using Delta Graph Context Representation”, Accepted by Journal of Systems and Software (JSS) Preprint
Services
I’m actively involved in the academic community, contributing my expertise in various capacities:
Conferences
- IEEE/ACM International Conference on Mining Software Repositories (MSR) 2024 (Junior PC)
- SANER Industry Track
Journal
- Applied Soft Computing Journal
Volunteer
- ICSE 2025, CSER 2025
Subreviewer
- ICSE, ASE, ICSA, SCAM, FSE, ICPC, SANER, SCAM, MSR, TOSEM (2024-2026)
Contact
first_name(at)dal.ca to get in touch for collaborating, or if you have any opportunity that suit’s my profile.