Computer Vision Based VR Simulation Project
Research Assistant, Advisor: Prof. and Dean. Peisen Huang
Hi, you found me.
I'm Yigao Fang (方益高/ほういこう/방이고), and I am currently interning as an associate quant developer at Barclays.
I am scheduled to graduate in December 2023 and I am actively seeking new graduate positions in the field of quant finance.
I possess strong skills in Python, C/C++, Java, and JavaScript. I have gained valuable experience in developing low-latency complex event processing (CEP) trading systems, as well as creating trading strategies such as smart order routing, specifically for market making and electronic trading in the Fixed Income business.
Additionally, I am proficient in working with Linux/Unix operating systems and have a solid understanding of machine learning concepts.
My academic resume can be found in the following link,
but I want to record more abundant myself here.
Now it's my story.
Implemented low-latency complex event processing (CEP) software trading systems such as algorithmic trading BARX Rates One
platform in Java and Apache Maven for market making and electronic trading in the Fixed Income business.
Developed advanced quantitative finance trading strategies of smart order routing (SOR) for Rates & Credit, and modeled 21
system behaviors of the Fill and Store limit orders with algorithmic realistic engineering (ARE) frameworks and QCC tests.
I designed a pipeline that preprocesses an open database and trains our improved deep learning model (GemNet+). With the trained model, I developed a platform that predicts force and energy of a catalyst based on structure using Python (Mean Avg. Error less than 0.30 eV).
I enjoyed the time when I was working with Haoyu Huang as a Teaching Assistant in Electromagnetics. We held Recitation Classes and Office Hours every week. I also helped Prof. Chen design and grade students' homework and exam papers, and posted solutions on my own.
As a team of five, we designed a horror game, Asylum 7 You can Play the Game. (Note: If you notice any issues, feel free to shot me an email. I appreciate your testing process!)
I design a new platformer game, Momory using Unity and C#, with a new mechanic of box collection and building system. You can Play the Game. (Note: If you notice any issues, feel free to shot me an email. I appreciate your testing process!)
I was employed as a grader for this course, working for more than ten hours per week to grade students' homework, in the topic of machine learning for computer vision. EECS 598-006 is a graduate-level course.
I reproduced The Legend of Zelda (1986) using Unity and C#. You can Play the Game. (Note: If you notice any issues, feel free to shot me an email. I appreciate your testing process!)
Starcli is an open source GitHub project, with the ability to browse trending projects on GitHub from the command line. I solved two issues of this project using the software development knowledge learned from the course. I set up the development environment on my local device and resolved the issues of adding filters with multiple languages and adding graph feature to show repositories. Finially, I opened two pull requests.
I designed a self-aggrandized encoding algorithm based on De bruijn series to generate grid patterns, and experimented the model by mounting camera on the VR helmet to capture grid patterns on the surrounding walls. I then implemented the decoding process to calculate the six-degree-of-freedom coordinates of the VR helmet using images captured by camera. Finally we realized a computer vision based VR simulation program with high accuracy to the scale of 1 mm.
I analyzed the audio of birdcalls and turn it into a spectrum map. Then I developed machine learning algorithms to identify bird sounds in continuous soundscape data, especially designed CNN networks to make identification of high correctness.
I applied machine learning methods to sort through and extract features of large data sets containing characteristics of investors, and eliminated redundant data and identified features of venture capitalists by making improvements to the existing Convolutional Neural Networks(CNN).
I used Multiple Object Tracking (MOT) techniques to extract the images of pedestrian from video clips captured by on-board cameras, and apply Computer Vision techniques to preprocess the images. Then the Net18 CNN model and PyTorch based LSTM pipeline are designed and applied to estimate the probability of pedestrian intention of crossing the road, making it easier for autonomous driving cars to avoid pedestrians.
I enjoyed the time when I was working with Juechu Dong and Haoyang Zhang as a Teaching Assistant in Honors Physics. We held Recitation Classes and Office Hours every week, and graded students' homework and exam papers.
We designed the best 3-dimensional geometric shape and optimal sand-to-water mixture proportion to use as a sandcastle foundation that will last the longest period of time on a seashore. We also simulated cellular automata and Smoothed Particle Hydrodynamics (SPH) model with C++ and visualized the simulation through ParaView. We are awarded as Meritorious Winner.
During the winter, I participated in the winter exchange program at the Sophia University in Tokyo. I took the Japanese Economy and Culture courses and spoke Japanese to native speakers. It is my first experience to study abroad in a foreign country. It was exciting to make friends with international students from Australia, Unite States, and Europe.
We designed a roller coaster and visualized it with 3D model, and quantitatively measure the safety and irritability of it. Euler's method was applied to track the motion of it at every instant of time. We are awarded as Golden Medal.
I studied and practiced three practical and effective anonymization methods, including k-anonymity, l-diversity, and t-closeness. Then K-means clustering method is applied as the core of our algorithm to effectively fulfill the k-anonymity and introduced utilize sensitive information as a new dimension to achieve l-diversity. To optimize the data points distribution, I introduced dichotomy methods and accomplished one report individually.
As a director, scriptwriter and actor, I participated in the rehearsal and performance of almost every drama since Oct. 2018. In particular, I was the leader of dedicated screenwriter group, executive director and actor of The Devotion of Suspect X, the assistant director of Hedda Gabler, etc. I was selected as the Club President due to my contribution.
I participated in the Club Department in the Student Union, and was selected as the Department Minister due to my contribution and leadership ability. We organized events such as Freshman Debate Competition, Undergraduate Knowledge Contest, etc. As a member of the Student Union, I actively guide freshmen to adapt to college life, and enriched the extracurricular activities of the institute.
Research Assistant, Advisor: Prof. and Dean. Peisen Huang
EECS442: Computer Vision
EECS445: Machine Learning
EECS401: Probability
University Physics Competition
Mathematical Contest in Modeling
Where then is the purpose,
which makes patient learning supportable,
when the world is on fire?
---- Kingman Brewster, Jr