Sep 2013 – Sep 2014 · 1 yr · London Area, United Kingdom

• Contributed ideas, code and data science to team effort on Bing AutoSuggest (search query formulation and auto-completion).
• Evangelized deep learning techniques for natural language processing at Bing Europe.
• Initiated a novel prototype for query formulation on Bing by fostering collaboration with Microsoft Research and gathering peer support; developed production-quality software for deep learning, evaluated it on Bing data and demonstrated to the management chain up to the executive VP that this R&D prototype achieved competitive metrics. Bing Europe winner and Bing finalist at the Microsoft-wide hackathon.

Jan 2011 – Aug 2013 · 2 yrs 7 mos · Murray Hill, NJ

• Conducted autonomous research on big data within the Statistics and Learning Research Department.
• Technical lead on indoor geo-location project. Developed and filed four patents on probabilistic WiFi and computer vision-based localization. Built robots for autonomous RF data collection; researched and hacked end-to-end software for RGB-D Simultaneous Localization and Mapping (SLAM). Shipped and coordinated robot deployment at engineering centers. Created Android apps for WiFi/LTE SLAM from the pocket. Enabled three major client projects for WiFi geo-location and the optimization of LTE network deployment.
• Technical lead on smart grid client project. Extensively investigated time series models (kernel methods, state-space models, sparse coding) for predicting electric usage, from individual meters level to citywide, achieving state-of-the-art load-forecasting results. Validated and delivered load-forecasting software solution to utility.
• Deployed Hadoop computing servers for the lab. Recruited and supervised four interns.

May 2010 – Jul 2010 · 2 mos · Florham Park, NJ

• Enhanced the state-of-the-art of speech recognition by developing a new statistical language model, validated on large text corpora (best known performance), filed a patent.

May 2009 – Aug 2009 · 3 mos · New York, NY

• Engineered end-to-end information extraction software for natural language processing and sentiment analysis of financial news. Researched deep learning algorithms for dynamic text topic modeling.

May 2008 – Aug 2008 · 3 mos · New York, NY

• Developed parallel (Map-Reduce) probabilistic graphical models for collaborative filtering (recommender system). Worked directly on the C++ production code in engineering team.

May 2006 – Dec 2008 · 2 yrs 7 mos

• Invented a breakthrough neural network algorithm for epileptic seizure prediction from EEG. Patented and published my software and research, conducted clinical case studies.

Sep 2005 – Dec 2010 · 5 yrs 3 mos · New York, NY

• Courant Institute of Mathematical Sciences.
• PhD advisor: Prof. Yann LeCun.
• Graduate teaching assistant:
• UNIX (Fall 2005),
• Machine Learning (Fall 2006),
• Compilers (Spring 2007).
• Tutor:
• C++ (Spring 2009), Scientific Calculus (Fall 2008), Computational Finance (Fall 2008),
• Digital Signal Processing (Fall 2007).

Aug 2003 – Jul 2005 · 1 yr 11 mos

• Schlumberger-Doll Research, Ridgefield, CT.
• Solved geology-related problems using computer science and mathematics, conducted client case studies.
• Patented algorithm for geostatistical software.
• Developed 3D models of fluid flow, software for automated image recognition on borehole image logs, well-to-well correlation and wavelet analysis of log curves, image processing for multipoint geostatistics.
• Successfully transferred research to engineering.

Jul 2002 – Aug 2003 · 1 yr 1 mo

• Schlumberger Cambridge Research, Cambridge, England.
• Patented neural networks software for geological classification.
• Conducted client case study.

Jun 2001 – Jun 2002 · 1 yr · Clamart, France

• Engineered two geophysical applications on a commercial platform.

Dec 1999 – Dec 2000 · 1 yr · Toulouse, France

• Negotiated contracts with IT business professionals, wrote technical specifications, managed projects. Designed image processing software for a French horse racing company.