- Gan Tu :: CV


Univerity of California, Berkeley

Computer Science Honors, Entrepreneurship & Technology - 3.925 GPA

  • 1 of 49 current honors students in the undergraduate computer science department
  • Academic Highlight: Dean's Honors; High A+ and Ranked 17 out of 800-1000+ students for CS61A/CS61B.
  • Honors Highlight: 8+ competition winners in areas including hackathon, engineering, entrepreneurship, consulting, and academics; Multiple prestigious externship and international conference participants.
  • Relevant Courses: Artificial Intelligence, Machine Learning, Deep Learning, Natural Language Processing, Data Structures, Machine Structures, Algorithms, Linear Algebra, Discrete Math, Multivariable Calculus, Statistics, Probability Theory, Signals, Circuits, Database, UI/UX Research

Univerity of Pennsylvania - The Wharton School

Business Specializations, Entrepreneurship Specializations - Online

  • Graduate of Wharton School of Business's online specialization programs in "Business Foundations" "Business Analytics," and "Entrepreneurship."
  • Using 2 years, I completed 12 classes with topics in: Marketing, Operations, Accounting, Finance, People Analytics, Customer Analytics, Operations Analytics, Accounting Analytics, Idea Conception, Product Design, Financing, Organizational Structure, and Management of New Enterprises.
  • Certifications and work samples available upon request.

ArtCenter College of Design

Graphic Design, Product Design, Transportation Design - ArtCenter at Night Program

  • Lead and completed graphic design projects (for-cause posters, identity systems, typographic counterparts) for various client needs. Designed personal medication 3d printing product concepts and concept car for McLaren.
  • Staff recognized student with portfolio samples available.

Profesional Profile

I am a Neo-Generalist, who works to become both a polymath and specialist.

I will be one of the few people you will ever encounter in life who have an immense curiosity and is ultra passionate about many subjects. I’ve actively spent time to learn, acquire, and refine multiple interdisciplinary skills. I’ve participated in several award winning projects, recognized nationally and internationally, across different domains. I’ve also spent 1200+ hours on online and summer classes in the past 4 years in subjects I am interested in outside of school.

The main domains I am working on right now are (1) business (entrepreneurship & finance), (2) computer science (software engineering & data science & system design & machine and deep learning), (3) law (contract law, IP law), (4) design (graphic, industrial, film, architecture, interior, entertainment), and (5) philosophy (modern philosophy, logic, mind).

In design discipline, I’ve graduated from USC's 14-week Architecture seminar and have designed multiple department-recognized architecture pieces with physical models. I’ve done creative advertising, for cause posters, and identity system design projects for companies. I’ve led photography, videography, graphic and web design projects for many collegiate clubs. I was the dance lead for school modern dance team, performing at various competitions and events, including university centenary ceremonies. I am also a National Geographic member, a frequent symphony goer, and a merit level 10 pianist.

In law displace, I’ve assisted, in parts, the drafting and translation of construction business contracts, including an international business contract between Chinese and Thailand Government. I am also a student of Haas School of Business contract law class. I am also familiar with basic IP laws, including concepts like USPTO track, PPH track, Trademarks, Patent Search, etc.

Work Experience

May 2018 - December 2018

Incoming Software Engineering Intern

Apple, Intern

7-month internship working on Machine Learning, Deep Learning, and Natural Language Processing for Siri and other international technologies.

September 2016 - Present

Deep Learning Researcher & External Officer

Machine Learning at Berkeley, Full Time

  • Researched lightweight, real-time object detection architectures that reduce the network size of FCN-based SqueezeDet by 12.5% with only 2% loss in mAVP using partial-layer weight quantization.
  • Lead club external relations, graphic design, and marketing efforts. Worked on machine learning tutorials, which were all ranked top 3 on HackerNews and ranked #1 on Reddit ML sub-reddit.
  • Wrote and deployed stock scrapers in AWS to crawl 1M+ S&P 500 minute-by-minute trading data from stock charts and store them in NoSQL and SQL database, later used to train trading predictors.

June 2016 - Present

Full Stack Web Developer


  • Designed & deployed 6+ responsive web apps (Flask, Angular.js) full-stack for various client needs.
  • Reengineered NASA Starlight program’s legacy aerospace photon propulsion online modeling tool with a new file system, front-end interface, and more sophisticated modeling functionalities.

Project Experience

May 2017 - Present

Autonomous Vehicle

Computer Vision & Deep Learning

  • Experimented and trained various neural network models (end-to-end, CNN, FCN) that classify traffic signs at 94% mAVP, perform generalizable behavioral cloning on human driving behaviors, and conduct image segmentation of human figures, based directly from front-camera image stream.
  • Integrated traditional image analysis techniques using OpenCV. Developed high-accuracy pipelines to identify lane lines and curvature, and to detect and track vehicles under varied lightings.
  • Program Kalman filters and particle filters using sensor fusion and Markov localization to precisely determine the location of other vehicles on a map at single-digit centimeter-level accuracy.
  • Build both PID controllers and model predictive controllers for advanced vehicle actuation.
  • Construct finite state machines for path planning using data and model driven behavior predictions.
  • Learn more from my comprehensive write-ups and video demos at code.tugan.me/self-driving-car.

April 2017 - Present

Machine Linguists

Natural Language Processing

  • Engineered and trained a MEMM-based POS tagger (95.7% mAVP), a transition-based dependency parser (67% mAVP), and a pronominal coreference resolution parser (77%+ mAVP) with custom engineered features and self-implemented Markov models, and decoding & parsing algorithms.
  • Engineered and trained both LSTM-RNN-based and seq2seq-based neural network models for English-French machine translation and Simpsons television scripts, language generation tasks.

May 2017 - Present

Intelligent Bots


  • Designed and implemented ROS-based prototypes that manipulate simulated robotic arms with six degrees of freedom for collision avoiding object localization, relocation, and motion planning.
  • Built autonomous rover models for high fidelity environment mapping and rock sample extraction.

Technical Skills

Advanced, 2 Years

AI & Machine Learning & NLP 75%

Graduate of Udacity Deep Learning and Self Driving Car Programs. Completed graduate level classes in Machine Learning, Deep Learning, Deep Reinforcement Learning and Natural Language Processing from UC Berkeley, Stanford, and MIT. Familiar with Python, Java, C, C++, Tensorflow, Keras, PyTorch, OpenCV, AWS, Git, UNIX, NumPy, Matplotlib, Pandas, NLTK, Apache Spark, etc.

Advanced, 2 Years

Full Stack Web Development 68%

Completed 5 series of classes in full stack development. Have designed and deployed 6+ responsive web apps for various client needs. Familiar with HTML, CSS, JavaScript (jQuery, AJAX, Angular.js), SQL, NoSQL, Bootstrap, UI/UX, Flask, Heroku, etc.


"I had an opportunity to observing Gan during his excellent work on the VICI Labs WaterSeer Collider project. His approach was thorough, imaginative, and completely professional. I was especially impressed by the clarity, insight, and technical excellence of his final report, and in his engaging manner of explanation and communication with his teammates, and with the VICI Labs executives."

Don Zacherl, CEO of VICI Labs, T3 TigerTech

"Few people have the opportunity to work with a developer who is also a creative designer and an insightful business advisor, but I did when I worked with Michael. I had the pleasure of working with Michael in France in a multi-cultural startup environment, collaborating on our respective projects. Michael’s ability to juggle multiple projects was unlike any I’ve seen before and made a dramatic boast in our team productivity. As a team member and a leader, Michael earns my highest recommendation."

Timothy Keynes, Entrepreneur & 42 Student

"I had the opportunity to work with Tu Gan during our time at the European Innovation Academy in Nice, France. He possesses a keen eye for software and UI/UX design, as well as the ability to learn new skills quickly in order to solve difficult problems. What sets Tu Gan apart from his peers is his dedication to excel in both technical and business roles. I recommend Tu Gan without reservation."

Zach Richardson, Founder of Tyro, LLC