Raymond Yuan

Biography

I'm a Rice University recent graduate (December, 2018) with a Bachelor of Science in Computer Science. I'm passionate, hard working, and a quick study. I love learning, coffee, dogs, drawing, exploring, and reading! I'm very passionate about machine learning and its applications. This summer, I'm headed to Nuro, working as a software engineer on the machine learning team for my full time job in the future. Read on and follow me on my social media below!

My Resume

Email me at ray.yuan0@gmail.com
if you have any questions!
Click here to download my resume.

Technical Skills

Fluent languages

Python
Java
Wolfram Mathematica

Frameworks

tensorflow
keras
git
android
unity

Proficient languages

c
HTML
C#
Matlab

Hardware

Through-hole soldering
SMD Soldering
Basic PCB design
Arduino embedded systems

Check me out on github

Publications

Blogposts

Work Experience

Google

Software Engineer

May 2018 - August 2018 | Mountain View, CA

Developed a series of guides introducing key tf.keras and machine learning concepts and best practices of each. Addressed a wide array of problems from classification tasks to eager neural style transfer to deep reinforcement learning to advanced research problems for different user profiles. Solved these problems in innovative ways that illustrate best practices. Hosted on TensorFlow Medium and tensorflow.org (NST, A3C).
Identified and fixed usability issues within Keras by developing features or making the user experience more streamlined and understandable.

NeoSensory

Algorithms Engineer

May 2017 - August 2017 | Palo Alto, CA

Implemented audio preprocessing libraries in Tensorflow complete with end to end unit testing. Incorporated with phoneme classification model for increased accuracy.
Wrote deep auto-encoder audio to haptic conversion algorithm in Tensorflow and Keras to optimize algorithm. Increased reconstruction accuracy by 83%. Converted algorithm to Android to run inference in real time.
Created and wrote framework for realtime phoneme classification in python and Tensorflow. Implemented quantization and other techniques to allow phoneme classification model to run in realtime.
Featured in National Geographic for work on infant haptic environmental sound algorithm (to be released in 2018)

NeoSensory

Applications Engineer

May 2016 - January 2017 | Houston, TX

Developed algorithms for music “sensationalizer,” which included beat detection, adaptive quantization, Fourier transforms of music. Prototyped the algorithm in Python and wrote it in C for real-time application.
Built apps in Android Studio (using Java) and Unity3D (using C#), coded firmware in Arduino.
Performed scientific experiments to determine best implementation for applications, performed statistical analysis, wrote technical memos, and presented on technical projects. Presented prototype and design to Global Fortune 100 company in Japan.

Activities and such

Position	Description	Time Period
Rice CS Club: President	Create vision and direction for CS Club Make new ways of connecting with and optimizing the CS undergraduate experience at Rice University Lead and organize club to build better events, experiences, and opportunities	Spring 2018 - Current
Patel Deep Learning Lab: Research Associate	Reverse engineer coarse-grained architectural motifs found in biology and neuroscience to solve perceptual tasks such as action recognition from video. I've also been working on showing that deep reinforcement learning can be improved in training time and in score through these techniques.	Fall 2017 - Current
McMurty College: Academic Fellow	Selected based on proven academic achievement and demonstrated willingness to help fellow students. provide free academic assistance on a regular basis through advertised review sessions, “office hours” in the college commons, and individual tutoring by request.	Spring 2018 - Current
COMP 540 - Statistical Machine Learning: Teaching Assistant	TA for graduate course COMP 540, Statistical Machine Learning. Hold office hours and recitation sessions	Fall 2017 - Current
Rice CS Club: External Vice President	Control all communications both to and from outside companies and organizations Coordinate company workshops and info sessions on campus Building out the first iteration of a technology opportunities page to present Rice students with all the available technology opportunities	Spring 2017 - Spring 2018
Rice Ultimate Frisbee: Fitness Officer	Lead and organize entire team warm up every practice. Coordinate and design external practice workouts, routines, and drills.	Fall 2016 - Spring 2017

My Projects

Here are my projects from the past couple years, ranging from those on github, wolfram demonstrations, to class projects. Unfortunately, lots of them were developed in private repos, but enjoy some of these that weren't! :)

Classifying Toxic Comments

Classified wikipedia comments for six possible types of comment toxicity (toxic, severe_toxic, obscene, threat, insult, identity_hate: multilabel classification). Implemented in Keras using fasttext embeddings, bidirectional GRU, spatial dropout and 1D Convolution. I also used a meta-learner in a superlearn blending style to ensemble different models. Classified with 0.9855 ROC AUC loss.
Rice Self Driving Car

I created and tested several models to solve the problem of predicting steering angle purely from video input feed. The best model was a Vision + LSTM stacked model: The images were first preprocessed with a vision stack comprised of a 3D CNN with 5 fully connected layers. This was then stacked on an RNN. The presented model is based on three key components: a 3D-CNN, a loss function built to generalize, and temporal context.
Wolfram Demonstrations

Bringing my ideas to life through the power of Wolfram Mathematica. Check it out!
Spotify Artist Connector

Ever wanted to find how an artist's music might changes into another? This project finds the shortest path between artists and generates a playlist that best represents this change in music. Uses Spotify API to generate the graph. Check it out!
Motivate Me!

Here's my Hack Rice 2016 Hackathon project! It's a social, organizational, and motivational app that encourages you and others to work towards a shared goal though competition. Uses firebase api to dynamically update databases. Check it out!
Mathematica Independent Study Projects

Here are some of the math projects and their applications I've worked on. Ranges from poly-style 3D models of a fish and its movement using rotation and translation matrices to image encryption: encrypting an input string with a variable size matrix key and encrypting it in an image to an in-depth analysis and proofs of Laplace Transforms and Bessel Functions. Check it out!

Global Temperature Classification Snapshot

Here's a project I did using webgl globe and api calls to categorize weather and temperature.

Github Profile

My general github profile for past projects and source code to these projects.