Machine Learning Consultant
2016 - 2018Yieldmo- Developed an innovative classification algorithm, using online learning, to predict the advertisers’ conversion rates and automate bidding strategies.
Technologies: Models, Reinforcement Learning, Scikit-learn, SQL, Java, PythonData Scientist, Ad Serving Team
2014 - 2016Yieldmo- Worked on display advertising, applying machine learning and reinforcement learning techniques to the problem of predicting click-through-rate.
- Implemented a dynamic A/B testing platform that allowed thousands of A/B tests to run in parallel.
- Used Bayesian Statistics and Thompson Sampling to redesign second-price auction rules and A/B testing decisioning.
Technologies: Models, Reinforcement Learning, Scikit-learn, SQL, Java, PythonMachine Learning Consultant
2013 - 2014Catalant Technologies- Built the first version of the matching engine from scratch using graph theory. This system recommends newly posted projects to MBA students on the platform and vice-versa.
- Ran market analysis to identify lags between offer (projects) and demand (MBA students' expertises).
Technologies: Graph Theory, Recommendation Systems, SQL, R, PythonConsultant in the Statistical Learning Team
2011 - 20121000Mercis- Developed a system in R recommending the best marketing means to reach each new customer on the website.
- Conducted analysis to build customer groups using unsupervised and supervised learning techniques.
Technologies: Clustering, Principal Component Analysis (PCA), Unsupervised Learning, Scikit-learn, Python, RAssistant Trader and Structurer
2010 - 2011UBS Investment Bank- Built a prototype tool that dynamically rebuilt a collateral portfolio using the risk of default analysis.
- Assisted the secured funding team in creating new products by running forecast tests and coming up with.
- Completed ROI and value propositions.
- Priced primary/secondary quotes and hedged currency strategies.
Technologies: Time Series Analysis, Visual Basic for Applications (VBA), RAssistant Researcher
2010 - 2010Cornell University, CFEM- Implemented a multi-agent system based on Ask/Bid and PnL data analysis simulating a stable financial market.
- Worked under the supervision of Dr. Sasha Stoikov, Head of Research at CFEM.
Technologies: Reinforcement Learning, C++, Java