Shinyoung Kim

Education

2021-2027 (expected): BSc, Computer Science; Kookmin Unversity, Seoul, South Korea

Experience

Winter Research Program on Modular Stan 2022 Jan.-2022-March

Worked on developing algorithms that can explore a network of models written in Modular Stan. Wrote Haskell code to extend the modular stan library and presented our results to a research seminar @ Columbia University and Stan devs. slides
Stan Development Team (Stan, https://mc-stan.org/) - Development Team Member, 2020 Nov.-2022

Implemented automatic differentiation variational inference algorithms for Bayesian inference in Stan with multiple collaborators using C++. Other minor contributions, such as new math functions for the Stan Math Library and interface improvements have been submitted and merged. As part of the Stan community helped organize a community session and seminar on Simulation Based Calibration for 100 participants, successfully running a tutorial session and lectures.
NextOpt - Software Developer, 2020 Jan.-2022

Implemented and deployed a quantity forecast service based on Bayesian seasonal forecasting models. I worked with 4 other software developers and data scientists to create and develop purpose-specific forecast models employing various tools such as stepwise selection, FBProphet, and Stan.
Software Maestro (https://www.swmaestro.org/) - Participant, 2019 June-2019 Dec.

Participated in Government funded software development program, where I was part of a team which developed a video processing framework for online game footage. I utilized video processing and statistical analysis to provide automated coaching tips to players.

Publications

International

“Simulation-Based Calibration Checking for Bayesian Computation: The Choice of Test Quantities Shapes Sensitivity”, Martin Modrák, Angie H. Moon, Shinyoung Kim, Paul Bürkner, Niko Huurre, Kateřina Faltejsková, Andrew Gelman, Aki Vehtari, 2025, Bayesian Analysis https://doi.org/10.1214/23-BA1404
“Implementation of user playstyle coaching using video processing and statistical methods in league of legends”, Shinyoung Kim, Dohyeon Kim, HyungGeun Ahn, Byeongtae Ahn, 2020, published on Multimedia Tools and Applications (https://doi.org/10.1007/s11042-020-09413-4)

Domestic

“A study on the Failure Function of Logistics Industrial Equipment Using Bayesian Aggregation: An Application to Electric Forklift Engine Failure Rate”, Shinyoung Kim, Jinwoo Choi, Kwangshik Nam, Hyunji Moon, 2021.

Projects

Rat (https://github.com/bbbales2/regressions, slides) Rat is a probabilistic programming language focused on simplicity in writing hierarchical bayesian models. It uses the notion of subscripted variables combined with categorical data to elegantly express nested hierarchical models. The original library included a crude Pratt parser and a compiler for transpiling into a Python module. After multiple refactors(read as rewrites), Rat code is now generated and executed dynamically through Jax JIT.
TraceInspector (https://github.com/kookmin-sw/2026-capstone-02) My bachelor’s graduation project. Implements a simple abstract interpreter for a subset of Go and animates fixpoint computation on the program state as an interactive web interface.
SBC (https://github.com/hyunjimoon/SBC) A R package for Simulation Based Calibration on Stan and brms models in collaboration with Hyunji Moon, Martin Modrák, and Teemu Säilynoja. I wrote the initial version and helped maintain the package which contains various tools for running SBC and analyzing results.
Stanify (https://github.com/Data4DM/stanify) A framework for specifying stan models on top of Vensim system dynamics models. Made a Vensim-to-Stan transpiler and a domain specific language for bayesian probabilistic programming.

Skills

Programming: Go, Coq, Python, C++

Language skills : Korean(native), English(Fluent)