Hi, I'm Jason Liang. Currently, I'm the product manager of the Image and Video team and a senior full stack software engineer at Scale, an AI infrastructure company with a current valuation of $7B+. Previously, I worked as a software engineer on the Alexa Engine infrastructure team at Amazon Lab126. I graduated from MIT, where I majored in computer science and mathematics with a minor in economics. My tech-related interests are mainly related to distributed systems and machine learning, especially deep learning and NLP.
ResumeOutside of academics, I helped organize the Harvard-MIT Math Tournament (HMMT), was a member of MIT's Asian Dance Team, and was the Risk Manager, House Chair, and President of the MIT Chapter of Zeta Beta Tau.
As Scripts Director for HMMT, I wrote a problem database using the Django framework which allowed question writers to submit proposed problems for review. This database included features such as real-time LaTeX rendering and programmatic generation of contest PDFs.
I was a member of the Prime Suspects team at the inaugural NBA Hackathon in New York City in , where we received 3rd place out of 40+ competing teams. We introduced a new metric for defensive performance based on a player's relative position between his assignment, the basketball, and the basket.
For the final project in my Machine Learning class (6.867), my collaborators and I extended the Google Brain paper Learning to Protect Communications with Adversarial Neural Cryptography. We used TensorFlow to build a system of adversarial neural networks, one of which was trained to encrypt messages with a shared symmetric key (Alice), one of which was trained to decrypt messages with the shared symmetric key (Bob), and one of which was trained to decrypt messages without the key (Eve). We explored several extensions, including using shorter shared keys and producing a concrete implementation of the cryptosystem.
My partners and I attempted to reproduce the results of the paper “Let There Be Light!”, which introduces a deep learning-based approach to image colorization.
Check out my Github for more projects!