Anup Deb

Email  /  CV  /  LinkedIn  /  GitHub

• Developed a Python and R toolkit to generate sophisticated research reports across the firm, which became widely adopted within Equities.
• Onboarded crucial Thompson Financial holdings datasets for Equities quantitative researchers.
• Built a distributed data storage platform for researchers to store and retrieve data and results efficiently.

• Designed and implemented an asynchronous Python library to allow traders and quantitative researchers to query the firm-wide portfolio monitor programmatically.
• Implemented digitization of margin agreements with brokers into expression trees which could be easily understood and optimized.

• Designed and implemented a command line tool which allows quantitative researchers to programmatically query information about jobs on the Cloud platform, adopted by 30+ researchers internally.
• Developed a FUSE filesystem in C++ which represents Cloud tasks as a hierarchical filesystem.

• Implemented backend infrastructure and frontend console for viewing examples of test errors (false positives, false negatives) to evaluate their deep learning image classification models and suggest improvements.

• Designed and implemented an algorithm to determine the introductory segments of TV show episodes using a MapReduce pipeline that operates on over 600,000 episodes in the Google Play Movies and TV catalog.
• Developed ‘Skip Intro’ on the Play Movies and TV Android application.
• Produced statistical analysis and visualizations of intro timestamps as a function of episodic signals (audio, video, shot boundaries) to estimate accuracy and limitations of the algorithm.