I am a final year PhD student at the University of Pennsylvania in the Wharton Department of Statistics and Data Science. I am advised by Jason Altschuler and closely work with Sophie Yu and Shuangping Li. I received my B.S. in Statistics, as well as in Mathematical Sciences from Seoul National University.

I am interested in the theory and practice of differential privacy and continuous optimization. My recent work focuses on developing new frameworks in privacy accounting and the design and analysis of first-order methods. I also study statistical inference in structured settings, such as random graphs.

I am looking for industry and academic research positions starting in Summer 2027.

Email: jinhobok [at] upenn [dot] edu

Research

Can Rényi Accountants Beat Lossless Accountants for DP-SGD? [arXiv][code]
J. Bok

Detection of Local Geometry in Random Graphs: Information-theoretic and Computational Limits [arXiv]
(alph) J. Bok, S. Li, S. H. Yu

Learning Acceleration Algorithms for Fast Parametric Convex Optimization with Certified Robustness [arXiv][code]
R. Sambharya, J. Bok, N. Matni, G. Pappas

Optimized Methods for Composite Optimization: A Reduction Perspective [arXiv][MP]
J. Bok, J. M. Altschuler
Mathematical Programming: Series A, 2026.
(Preliminary conference version appeared as:
Accelerating Proximal Gradient Descent via Silver Stepsizes [arXiv][PMLR]
J. Bok, J. M. Altschuler
COLT 2025.)

Shifted Interpolation for Differential Privacy [arXiv][PMLR][code]
J. Bok, W. Su, J. M. Altschuler
ICML 2024
FORC 2024 (non-archival track).

On the Privacy of Noisy Stochastic Gradient Descent for Convex Optimization [SICOMP]
(alph) J. M. Altschuler, J. Bok, K. Talwar
SIAM Journal on Computing, 2024.
(Preliminary conference version appeared as:
Privacy of Noisy Stochastic Gradient Descent: More Iterations without More Privacy Loss [arXiv][NeurIPS]
J. M. Altschuler, K. Talwar
NeurIPS 2022.)