|
Sepanta Zeighami
Postdoc at UC Berkeley zeighami@berkeley.edu Google Scholar | GitHub | LinkedIn |
Bio
I'm a Postdoctoral Scholar at University of California, Berkeley, advised by Aditya Parameswaran. Before that, I was a PhD student at USC's Infolab, advised by Prof. Cyrus Shahabi, from Aug/2019 to Feb/2024.
My research broadly focuses on data and AI systems. I work on building accurate, reliable, and efficient data-centric systems, with a penchant for theoretically understanding the use of machine learning in such systems.
Selected Publications
-
Cut Costs, Not Accuracy: LLM-Powered Data Processing with Guarantees
S. Zeighami, S. Shankar and A. Parameswaran
Proceedings of the 2026 International Conference on Management of Data, SIGMOD '26.
Code -
NUDGE: Lightweight Non-Parametric Fine-Tuning of Embeddings for Retrieval
S. Zeighami, Z. Wellmer and A. Parameswaran
Proceedings of the 13th International Conference on Learning Representations, ICLR '25.
NUDGE is incorporated in LlamaIndex framework! ( learn how to use )
Code, Blog Post
-
LLM-Powered Proactive Data Systems
S. Zeighami, Y. Lin, S. Shankar and A. Parameswaran
IEEE Data Engineering Bulletin, Issue on LLMs-meets-data. March 2025.
-
Theoretical Analysis of Learned Database Operations under Distribution Shift through Distribution Learnability
S. Zeighami and C. Shahabi
Proceedings of the 41st International Conference on Machine Learning, ICML '24.
Oral Presentation (1.5% oral acceptance rate from 10k submissions)
-
Towards Establishing Guaranteed Error for Learned Database Operations
S. Zeighami and C. Shahabi
Proceedings of the 12th International Conference on Learning Representations, ICLR '24.
-
BiasBuster: a Neural Approach for Accurate Estimation of Population Statistics using Biased Location Data
S. Zeighami and C. Shahabi
Proceedings of the 25th Conference on Mobile Data Management, MDM '24 .
Best Paper Runner-Up
-
On Distribution Dependent Sub-Logarithmic Query Time of Learned Indexing
S. Zeighami and C. Shahabi
Proceedings of the 40th International Conference on Machine Learning, ICML '23.
-
NeuroSketch: Fast and Approximate Evaluation of Range Aggregate Queries with Neural Networks
S. Zeighami, C. Shahabi and V. Sharan
Proceedings of the 2023 International Conference on Management of Data, SIGMOD '23.
Code.
-
A Neural Approach to Spatio-Temporal Data Release with User-Level Differential Privacy
R. Ahuja*, S. Zeighami*, G. Ghinita, and C. Shahabi. * Equal contribution
Proceedings of the 2023 International Conference on Management of Data, SIGMOD '23.
Code.
-
A Neural Database for Answering Aggregate Queries on Incomplete Relational Data
S. Zeighami, R. Seshadri, and C. Shahabi
Transactions on Knowledge and Data Engineering, TKDE '23.
Code.
-
A Neural Database for Differentially Private Spatial Range Queries
S. Zeighami*, R. Ahuja*, G. Ghinita, and C. Shahabi. * Equal contribution
Proceedings of the VLDB Endowment Volume 15, 2022.
Code.
-
Towards Accurate Spatiotemporal Covid-19 Risk Scores Using High Resolution Real-World Mobility Data
S. Rambhatla*, S. Zeighami*, K. Shahabi, C. Shahabi, and Y. Liu. * Equal contribution
ACM Transactions on Spatial Algorithms and Systems, 2022.
-
Estimating Spread of Contact-Based Contagions in a Population Through Sub-Sampling
S. Zeighami C. Shahabi, and J. Krumm
Proceedings of the VLDB Endowment Volume 14, 2021 .
Code.
-
Secure Dynamic Skyline Queries Using Result Materialization
S. Zeighami G. Ghinita, and C. Shahabi
2021 IEEE 37th International Conference on Data Engineering, ICDE '21.
Code.
-
Finding Average Regret Ratio Minimizing Set in Database
S. Zeighami and R. C. W. Wong
2019 IEEE 35th International Conference on Data Engineering, ICDE '19.
Code.
-
Minimizing Average Regret Ratio in Database
S. Zeighami and R. C. W. Wong
Proceedings of the 2016 International Conference on Management of Data, SIGMOD '16.
Code.