Computer Science Machine

Kui Tang

**** ****** **** ******.***/kuitang kui-tang.com

**** ******** ******@********.***

New York, NY 10027 319-***-****

Columbia University, New York, NY

Education

Bachelor of Science, Applied Mathematics Expected May 2014

• Advisors: Tony Jebara and Martha Kim.

• Relevant graduate courses: Convex Optimization Theory, Graph Theory, Distributed Systems,

Linear Programming, Measure-theoretic Probability, Bayesian Nonparametrics.

Large-scale learning, combinatorial optimization, distributed algorithms, Bayesian nonparametrics.

Interests

Computer Science Department, Columbia University, New York, NY

Research

Tractable Inference in Graphical Models 2013 – present

• Develop exact and provably approximate algorithms for Markov random fields.

• Develop graph-theoretic methods for privacy-aware data mining.

• Apply to energy and social network problems.

Bayesian Nonparametric Computational Neuroscience 2012 – 2013

• Automatically constructed 3D models of neural connectivity from electron micrograph images.

• Developed large-scale inference methods for dependent Dirichlet processes.

Parallel Computer Architecture and Compilers 2011 – 2012

• Developed a source-code based profiler for measuring scalability in multithreaded code.

• Applied machine learning methods and engineered features to identify bottlenecks.

Hunch, Inc., New York, NY

Industry

Machine Learning Intern 2011

• Developed a privacy-respecting revenue generating product based on learned user preferences.

• Evaluated collaborative filtering algorithms from the literature.

Publications

[1] K. Tang, A. Weller, T. Jebara. Network Ranking with Bethe Pseudomarginals. NIPS Workshop

on Discrete Optimization in Machine Learning. 2013.

[2] K. Choromanski, T. Jebara, K. Tang. Adaptive Anonymity via b-Matching. Neural Information

Processing Systems (NIPS). 2013. Spotlight accept rate: 3.7%.

[3] M. Kambadur, K. Tang, J. Lopez, and M. Kim. Parallel Scaling Properties from a Basic Block

View. International Conference on Measurement and Modeling of Computer Systems (SIGMET-

RICS) (Poster). 2013.

[4] M. Kambadur, K. Tang, and M. Kim. Collection, Analysis, and Uses of Parallel Block Vectors.

IEEE Micro 33(3):86-94 (2013). Top pick accept rate: 14%.

[5] M. Kambadur, K. Tang, and M. Kim. Harmony: Collection and Analysis of Parallel Block

Vectors. International Symposium for Computer Architecture (ISCA) 2012. Accept rate: 18%.

Invited Talks

[1] Adaptive Anonymity via b-Matching. Machine Learning and Friends Lunch. UMass Amherst,

Feb. 2014.

[2] Statistical Machine Learning with Bayesian Networks. Tutorial. Columbia Data Science Society.

Columbia University, Nov. 2013.

[3] Statistical Machine Learning with Bayesian Networks. Tutorial. hackNY Masters. New York

University, Sept. 2013.

Computer Science Department, Columbia University, New York, NY

Teaching

Teaching Assistant for Computer Science Theory 2011 – 2012

• Held recitation and office hours (3+ hours weekly) and graded homeworks and exams.

• Worked with class of 80 sophomore and juniors.

Youth for Debate, Columbia University, New York, NY.

Teacher 2010

• Taught public speaking and advocacy weekly to 30 high school juniors in a local public school.

• Coached students one-on-one on speech writing and performance.

Columbia University Egleston Research Fellowship ($10,000) 2010 – present

Awards &

Fellowships Runner-Up, CRA Outstanding Undergraduate Researcher 2014

Honorable Mention, CRA Outstanding Undergraduate Researcher 2013

NSF Research Experience for Undergraduates (REU) Fellowship ($7,000) 2013

UC Santa Cruz Machine Learning Summer School Scholarship ($725) 2012

NSF Research Experience for Undergraduates (REU) Fellowship ($8,000) 2012

hackNY Fellow ($4,500) 2011

Workflow Chair, International Conference on Machine Learning (ICML) 2013 – present

Service

Volunteer, Neural Information Processing Systems (NIPS) 2013

Co-organizer, Columbia Machine Learning Reading Group 2012 – present

President, Society for Industrial and Applied Mathematics (SIAM), Columbia 2012 – 2013

Treasurer, Beta Theta Pi 2011 – 2012

Committee Member, Association for Computing Machinery (ACM), Columbia 2011 – 2012

• MexCpp: Object-oriented C++ interface for writing MATLAB extensions (MEX) without tears.

Open Source

• Chauffeur: Drives machine learning experiments on clusters. Manages workflow and data.

• Harmony: Efficiently collect parallel block vectors (profiles) for multithreaded problems.

• French: Professional working proficiency (studied for 7 years in secondary school).

Languages

• Chinese (Mandarin): Native speaker.

• Fluent in C, C++, Python, MATLAB (including MEX), SQL, Java, L TEX, Shell.

A

Programming

• Erd s number is at most 4 (Tony Jebara Tommi Jaakkola Noga Alon Paul Erd s).

o o

Other

Last updated February 5, 2013.

Contact this candidate