David Suh の 일상.

M.S. Computer Science
B.S. {Mathematics, Computer Science (spec. ML)}
GPA: 3.79/4.0
Relevant Coursework: Honors (Real) Analysis, Probability Theory (acc.), Honors Algebra, Networks (graduate), Machine Learning (doctorate), Computer Vision, Honors Theory of Algorithms, Convex Optimization, Topics in SSL, Bayesian Classifiers for Finance.

Software Engineer Intern

• Incoming engineer to work on crypto and decentralized verification systems for Brave Software and their Brave products, in particular their video chatting product Brave Talk. Also working on the Brave Rewards team, which is responsible for the Brave Ads system and the Brave Wallet.

Machine Learning Engineer Contractor (prev. Intern)

• Developed a machine learning solution for a system that automatically detects, crops, and uploads student work directly from their webcam feed.
• Work included creating a docs-60k dataset of documents in natural backgrounds by curating from multiple existing datasets, fine tuning an instance of DeepLabV3 on PyTorch while using Weights and Biases for MLOps and optimization, and then preparing the model for edge device client-side deployment through quantization-aware training and using ONNX runtime web with React.js.

Research Assistant

• Worked as a research assistant in maintain and developing research related software and tools, including the maintenance and migration of the research group’s website using Django, react, and other data visualization tools such d3. Furthermore, implemented an optimized data pipeline uploaded papers and implemented an optimized segment tree system for searching.

Tutor/Grader

• Graded for Accelerated (Real) Analysis III, an advanced 2nd/3rd year math course at UChicago, as well as tutoring first-year Calculus.

1^st place team

• Came in first in the Midwest Regional for Citadel’s 2023 Datathon. Winning paper was written on the environmental (and consequently health-related) effects of greenspaces and bluespaces, in particular on surface temperatures and air quality index through programmatic time series analysis, using metrics such as the Dicky-Fuller test, and providing strong evidence of data correlation through K-means segmentation and heuristic minima discovery.

1000^th out of ~4,000

• Scored a 9/120^[1] on the 2022 Putnam Exam, a score which is just under the top 1000, about the 75th percentile overall.

Paid Participant

• Attended the UChicago Mathematics REU, a competitive program for undergraduates to spend 8 weeks emulating the graduate mathematics research process under the supervision of a mentor. The final paper, Mean Field Games and Weak Solutions, written on the topic of stochastic mean field games, a natural topic that forms from optimal control of the Fokker-Planck equation, which concerns the optimization of the actions of a small agent compared to a large population. In particular, we consider the system $$ \begin{cases} \partial_t u + H(x,D_xu, D_x^2u, m) = 0\\ \partial_t m + \operatorname{div}(H(D_pHm))-\partial_{ij}(D_{M_{ij}}H(D_pHm)) = 0\\ \end{cases} \quad \text{subject to} \quad \begin{cases} u(x,T) = \psi(x, \theta(\cdot, T))\\ m(x,0) = m_0(x) \end{cases} $$

Python (numpy/scipy, openCV2, cvxopt, pyTorch, TensorFlow, sklearn), C/C++, SQL, JavaScript/TypeScript, golang, R , java , Machine Learning, Convolutional Neural Networks, Graph Neural Networks, Computer Vision Systems

English (native), Korean (native), Mandarin (conversational)