Chu's Wine is built out of my passion to wines and the vision of all restaurants being able to offer successful wine programs, ones that not only generate profits but also delight their customer experiences, to their customers at absolutely Zero Cost.

Designed and developed a highly modularized & unified research/simulation/trading system that allows portfolio managers to rapidly iterate through ideation->prototype->simulation->live trading cycle with nearly zero quant dev involvement for productionization.My novel design elegantly solves the traditional dilemma of optimizing for high throughput required for research & simulation vs optimizing for low latency required for live trading. Furthermore, the system is highly modularized, to allow portfolio managers to focus their research & simulate upon specific component one at a time, e.g. alpha signal, portfolio optimization, forecast, etc. or end-to-end across any combination of components.

Lead and manage the MacroPipeline engineer team for designing and implementing the next generation of research/simulation/trading infrastructure for the firm. The infrastructure is optimized toward trading futures intraday and the main objective is to minimize the time to deliver from ideation to trading live.Key result includes- Design and lead the implementation of the new spark based research infrastructure which achieves at least 100X performance improvement in various critical research tasks including ticks aggregation, signals calculation, model fitting, and L2 tick simulation- Unify the research framework and the living trading system to share the same interface which allows researchers to deploy research code to live trading system as is- Develop the engineering driven culture and the corresponding discipline and best practice for the team.

Our fundamental believe is that resumes are biased and noisy. With resumes being the top of the funnel of hiring process, many qualified candidates who don't have strong resumes are not getting the opportunity they deserve and vice versa. My role at OneInterview includes product design, infrastructure design and development of our data-driven online coding assessment platform which allows software engineers to show rather than tell their skill set. Our pilot program with Baidu Research USA had helped double their hiring throughput with 5X better hiring conversion rate than candidates from other channel from which candidates did not going through the assessment from OneInterview.

Codecademy provides a cloud-based, virtualized way for users to self-teach computer programming by writing and running code directly into a console that lives inside a Web browser window. It is completely platform independent, turn-key and self-paced, and the usage data and learning outcomes feedback to the designing of the delivery and content. At Codecademy I led all data related efforts including the modeling and infrastructure behind- the machine learning models for studying user engagement and learning outcomes- the monitoring tools, statistical tests and analyses for steering product development based on usage data- the tools and infrastructure for business intelligence and analytics- the tools for monitoring and engaging in "growth hacking" (e.g. efforts to engineer viral growth in user base)I also designed and implemented an EventHub system for tracking and funneling users and for cohort analysis

Completely revamped and reengineered Square's fraud detection system- Designed and built the internal fraud detection machine learning pipeline in Python.- Designed and built the internal risk engine and signal service in Java- Designed and prototyped the merchant search system in Rails + AngularJs + Krati (Java-based search system based on forward index + bloomfilter rejection rather than inverted index)- Designed and prototyped the initial merchant analytics system in R

Established and headed LinkedIn's search relevance unit- Designed and deployed the first LinkedIn people search relevance model- Designed and prototyped the first version of universal search backend in Rails- Designed and implemented the internal L1 and L2 regularized multi-class logistic regression optimizer with cumulative sub-gradient descent in Java

Built a full stack of backend systems, website and mobile application for eCommerce startup

Led the link analysis fraud detection infrastructure underlying Google's Checkout and AdWord businesses20% project on implementing a few onebox widgets for Google mobile search frontend 20% project on implementing the distance meter for Google map frontend (My Map)