Jaehyun Park
 |
Jaehyun Park, Ph.D. |
Education
Ph.D., Computer Science, Stanford University, 2012–2017
Co-advised by Professor Stephen Boyd, and Professor Leonidas Guibas.
Dissertation: Approximation techniques for mixed-integer quadratic problems
B.S. with distinction, Computer Science, Stanford University, 2008–2012
Full financial support by Samsung Scholarship, 2008–2017
Work
Research scientist at Facebook, Ads Core ML team, 2017–2021
Tech lead of a 4-engineer team. Solves problems in the intersection of ML and systems.
Developing a feature exploration platform to enable quick experimentation with new feature ideas.
Acts as a strategist to formulate delivery paths and backup plans for projects depending on unbuilt infra capabilities.
Developed alternative measurement and roll-out methods to enable quick data and signal iterations for ML.
Built a real-time platform that collects large-scale signals, extracts semantic information, and aggregates them into user profiles for personalized ads across Facebook.
Developed a user-ad fuzzy matching algorithm to improve semantic relevance.
Ph.D. research intern at Instagram ads delivery, 2016
Ph.D. research intern at Microsoft Research (advised by Professor Sham Kakade), 2013
Software engineer at imo.im, 2011
Software engineer at Facebook, 2009
Research
Jaehyun Park, “Approximation techniques for mixed-integer quadratic problems”, Ph.D. dissertation, Stanford University, 2017. (external link)
Jaehyun Park and Stephen Boyd, “General Heuristics for Nonconvex Quadratically Constrained Quadratic Programming”, arXiv:1703.07870, 2017. (external link, QCQP Python package repository)
Jaehyun Park and Stephen Boyd, “A Semidefinite Programming Method for Integer Convex Quadratic Minimization”, Optimization Letters, 2017. (external link)
Jaehyun Park, “Sparsity-Preserving Difference of Positive Semidefinite Matrix Representation of Indefinite Matrices”, arXiv:1609.06762, 2016. (pdf)
Jaehyun Park and Stephen Boyd, “Concave Quadratic Cuts for Mixed-Integer Quadratic Problems”, arXiv:1510.06421, 2015. (external link)
Jaehyun Park and Maurizio Calo, “Symbolic Subdifferentiation in Python”, Independent work (under the supervision of Professor Stephen Boyd), 2011. (pdf, codes)
Jaehyun Park, “Recognizing Human Actions in Videos”, Stanford CURIS project (under the supervision of Professor Andrew Ng), 2010. (slides)
Jaehyun Park and David Kamm, “Cortically Inspired Architectures for Action Recognition in Movie Clips”, Independent work, 2010. (pdf)
Michael Harris, David Kamm, Jaehyun Park, and Jeffrey Wang, “Unsupervised Learning of Invariances in Deep Networks”, Independent work (under the supervision of Professor Andrew Ng), 2010. (slides)
Michael Harris, Jaehyun Park, and Jeffrey Wang, “Deep Networks on the GPU”, Independent work (under the supervision of Professor Andrew Ng), 2010. (pdf)
Jaehyun Park, “Clone Detection in a Software Using Graph Modeling”, Independent work (under the supervision of Professor Byung-Ro Moon), 2006. (pdf in Korean)
Awards
ACM International Collegiate Programming Contest (ACM-ICPC)
World Finals 2010: North America Champions (14th place)
World Finals 2009: 20th place
Pacific Northwest Regional 2009: 2nd place
Pacific Northwest Regional 2008: 1st place
International Olympiad in Informatics (IOI)
2006: Gold Medal (2nd place worldwide)
Asia Pacific Informatics Olympiad (APIO)
2007: Silver (7th place)
Teaching
Convex Optimization Short Course at ShanghaiTech University, co-instructor, March 2016
EE 364a (Convex Optimization I) course assistance, 2013–14 summer
EE 364b (Convex Optimization II) course assistance, 2013–14 spring
EE 364a (Convex Optimization I) course assistance, 2013–14 winter
CS 161 (Design and Analysis of Algorithms) course assistance, 2012–13 spring
CS 97SI (Introduction to Programming Contests) instructor, 2011–12 winter
ACM-ICPC Pacific Northwest problem setter and judge
Stanford ACM-ICPC coach and problem setter, 2010–2015
ACM-ICPC related materials on GitHub.
Korea Olympiad in Informatics (KOI) coach and problem setter, 2008, 2012
International Olympiad in Informatics (IOI) student coach of the Korea team, 2013
Korea team won 2 golds and 2 silvers, placing 3rd (unofficial country rank).
USA Computing Olympiad (USACO) coach and problem setter, 2008–2011
Some old USACO stuff can be found here.
Piano
Chopin, Nocturne in E-flat (May 2016): video
Chopin, Nocturne in B (March 2016): video
Liszt, Transcendental Etude 'Feux follets’ (December 2015): video
Gershwin-Grainger, 'Love Walked In,’ with Chopin's nouvelles etude #2 as a prelude (November 2015): video
Gershwin, Rhapsody in Blue (first piano: Patrick Kim, June 2015): video
Schubert-Liszt, Ständchen (June 2015): video
Chopin, Barcarolle in F sharp (June 2014): video
Kapustin, Concert Etude 'Reverie’ (June 2014): video
Kapustin, Concert Etude 'Intermezzo’ (March 2014): video
Schumann-Liszt, Widmung (December 2013): video
Kapustin, Concert Etude 'Toccatina’ (December 2013): video
Levitzki, Waltz (November 2013): video
Kapustin, Sonata No. 2, Movement 1 (June 2013): video
Kapustin, Variations (November 2012): video
Jaehyun's Senior Recital live recordings (May 2012)
Liszt, Ballade No. 2 in B minor: video
Chopin, Ballade No. 4 in F minor: video
I won the Humanities & Sciences Undergraduate Prize in Music for my performance of Chopin's 4th Ballade.
Kreisler-Rachmaninoff, Liebesfreud: video
Kreisler-Rachmaninoff, Liebesleid: video
Gershwin, Rhapsody in Blue: video
Bolcom, Graceful Ghost Rag: video
Beethoven, Cello Sonata No. 3 in A major, Movement 1 (cello: John Pardon, February 2012): video
Miscellaneous
-
This video was part of a PandoDaily article “Top 5 Stanford CS Students You Should Know,” posted on March 30th, 2012. The article is gone now, and only the video is remaining.
-
Peking University Online Judge Member (ID: rouder)
I used to solve problems on this website for my IOI and ICPC training. It's outdated now, and I recommend Codeforces or topcoder for people who want to improve their algorithmic problem solving skills.
Random codes I wrote
Simple VLC extension for cropping videos using ffmpeg.
Python script for downloading Twitch.tv VODs. This needs ffmpeg to be in PATH, and the m3u8 package to be installed.
