CV
General Information
| Full Name | Shristi Das Biswas |
| sdasbisw@purdue.edu | |
| GitHub | ShristiDasBiswas |
| Google Scholar | Profile |
| Profile | |
| Languages | English, Bengali, Hindi, Japanese |
Education
-
2021 – Present PhD in Electrical and Computer Engineering
Purdue University, West Lafayette, Indiana, USA - Advisor: Prof. Kaushik Roy
- Research Topic: Energy-Efficient Deep Learning and Responsible Generative AI
- GPA: 3.83 / 4.0
-
2017 – 2021 B.Tech. in Electronics and Telecommunication Engineering
Indian Institute of Engineering Science and Technology (IIEST), Shibpur, India - GPA: 9.39 / 10.0 (Department Rank 2 / 61)
Internship Experience
-
2025 – 2025 Applied Science Intern
Amazon Kumo (AWS Support), Bellevue, WA - Developed continual learning algorithms for large language models (ELLA), reducing catastrophic forgetting while improving forward transfer and scalability.
- Achieved up to 9.6% relative accuracy gain and 35× smaller memory footprint across LLaMA and T5 models with formal analysis of anisotropic shrinkage regularization.
- Benchmarked against Claude 3.5 Haiku/Sonnet showing ∼ 59× lower latency and 99% lower annual cost.
- Workshop Paper accepted at NeurIPS CCFM Worshop 2025; Full paper submitted to ARR.
-
2024 – 2024 Applied Science Intern
Amazon Fashion, Sunnyvale, CA - Designed PIXELS, a progressive exemplar-driven image editing framework using diffusion models for customizable and imitation-based editing during inference.
- Improved exemplar fidelity by 41% and visual realism by 33.6% in user studies.
- Full paper accepted at AAAI 2024.
Honors and Awards
-
2025 NeurIPS 2025 Scholar Award
- Awarded for research on efficient and responsible AI systems in diffusion and LLMs.
-
2023 Google CS Research Mentorship Program Scholar
- Selected for Google CS Research Mentorship Program 2023b Cohort.
-
2020 Indian Academy of Sciences Summer Research Fellowship
- Selected for competitive national research program by IAS.
-
2020 IEEE Women in Engineering Best Student/Research Scholar Paper Award
- Awarded for paper presented at ICDCN 2020.
-
2019 Indian Youth Delegate to China
- Selected nationally by Ministry of Youth Affairs and Sports, Government of India.
Academic Service
- Reviewer for Conferences: NeurIPS, ICLR, CVPR, ICCV, AAAI, WACV, IJCNN
- Program Committee: AAAI
Technical Strengths
- Programming Languages: Python, MATLAB, C/C++, Bash
- Frameworks: PyTorch, TensorFlow, LaTeX
- EDA and Hardware Tools: Cadence Virtuoso, LTSpice
- Version Control: Git, GitHub
- Languages: English (fluent), Bengali (native), Hindi (proficient)