top of page
danny_profile_pic.jpg

I'm interested in how data, probability, and computation can be used to make better decisions under uncertainty. My work so far spans stochastic modeling, inference for partially observed Markov process (POMP) models, and applications in finance and climate-health

I’m building toward a career at the interface of modeling and decision-making: designing algorithms and systems that turn noisy real-world data into reliable, impactful decisions.

About Dae Hyun(Danny) Kim

Education

University of Michigan, Ann Arbor, MI | Aug 2021 – May 2025

BS, Statistics (Honors) & Mathematics | Graduation Date: May 2025

Programming Languages & Tools: Python, C++, SQL, R, MATLAB, Mathematica, LaTeX, Git, HTML, Tableau

 

Libraries: JAX, TensorFlow, PyTorch, Keras, NumPy, SciPy, scikit-learn, Pandas, Matplotlib, NetworkX

Research

SPSA - Optimized Ranker Ensemble for Movie Recommendations

Advisor: Serin Hong

  • Designed and implemented a movie recommender algorithm that directly optimizes top-K quality using Simultaneous Perturbation Stochastic Approximation (SPSA)

  • Benchmarked standard recommenders with HitRate@K on 200k+ users (32M+ ratings); validated with user-level paired bootstrap.

Hyperbolicity of Fine Arc Graphs

Advisor: Roberta Shapiro

  • Conducted in-depth study of geometric group theory and topological graph structures to investigate the hyperbolicity of fine arc graphs on compact orientable surfaces.

  • Developed a proof outline, using surgery arguments and quasi-isometry frameworks.

Undergraduate Honors Thesis

Advisor: Edward Ionides

  • Debugged pypomp and automatic differentiation-based algorithms using the Heston stochastic volatility model under JAX, which was for Thesis, Testing a Modern Inference Framework for POMP Models: A Case Study Using Stochastic Volatility

  • Designed Monte Carlo experiments and diagnostic tools to compare pypomp against R’s pomp, ensuring correctness and reproducibility.

  • Paper | Github

Python Package Development for POMP Models

Advisor: Edward Ionides

  • Implemented Monte Carlo-Adjusted Profile Likelihood (MCAP) for pypomp,  a JAX based Python package for inference on partially observed Markov process (POMP) models, constructing profile-likelihood confidence intervals that separate statistical and Monte-Carlo error, enabling reliable uncertainty quantification for stochastic, simulation-based inference.

  • Github

Evaluating DEI Impact Using Synthetic Control Method

Advisor: Yuekai Sun

  • Conducted a causal inference study to evaluate the impact of the University of Michigan’s Diversity, Equity, and Inclusion (DEI) program using the Synthetic Control Method.

  • Demonstrated robust positive treatment effects for DEI across racial and economic access metrics, reinforcing causal links between policy implementation and improved student equity outcomes.

Work Experience

DYD EDU Co., Ltd.

ARK Impact Asset Management Inc.

Department of Mathematics, University of Michigan

Data Scientist

Intern

Teaching Assistant & Grader

  • Built POMP models that integrate satellite AOD and atmospheric reanalysis to forecast exceedance probabilities for WBGT and PM2.5, enabling risk guidance

  • Turned probabilistic forecasts into activity decisions for schools and local health agencies

  • Stabilized training under noisy likelihoods using stochastic optimization with adaptive step sizes and GPU acceleration, allowing higher particle counts and more reliable tail estimates

  • Contributed time-varying dependence modeling (BTC-S&P 500), using DCC-GARCH, analyzing correlations spikes during market stress 

  • Researched Impact Investing Determination to detect startups with huge potentials for investment decisions 

  • Worked for stablecoin policy research project; analyzed risk controls and e-commerce impacts on banks/card networks 

  • Provided open sessions to 60+ students (two consecutive semesters) for proof techniques & mathematical logic 

  • Contributed to developing students’ quantitative reasoning and improving academic achievement, grown by 14.8% 

  • Organized quiz & exam prep sessions for linear algebra students; assisted the professor with grading students’ papers 

ROK Army (Korea Combat Training Center)

Military Service: Sergeant

  • Led 150+ platoon members during PT & FTX drills; received an honorable discharge in recognition of top-most service 

  • Contributed to war simulation data analysis using the Multiple Integrated Laser Engagement System (MILES) 

  • Provided 50+ summer & winter data analysis boot camps for Supply Chain Optimization (SCO) & cybersecurity 

  • Training Videos 1, 2

Hobbies & Interests

Outside of academics and research, I try to keep my life active and curious:

  • Basketball - I enjoy practicing dribbling drills and playing pickup games. Also, a big fan of NBA! It is a way to reset my mind!

  • Running - It reinforces STAMINA in studying and research! Showing up consistently, staying focused through long stretches, and pushing through moments when progress is slow!

  • Traveling - I like exploring new cities and cultures. It often gives me new perspectives on the kinds of problems I want to work on.

bottom of page