Benjamin Breton Email & Phone Number

New York City Metropolitan Area

Instagram Graph Ranking

United States

I started working as a freelance data scientist in July 2017 in tandem with my full-time jobs. Freelancing allows me to work with different technological stacks and expand my expertise.I am passionate about the possibility of working across multiple industries to identify unique technological needs and determine how I can use data to increase company value.I have thus far had the opportunity to collaborate with 8 companies on 18 projects. More details are available on my Malt profile (in French).𝐃𝐚𝐭𝐚 𝐚𝐧𝐚𝐥𝐲𝐬𝐢𝐬- Performed statistical analysis on tweets, online reviews, resumés, navigation data and customer data, using databases such as MYSQL, POSTRGE and MONGODB.- Designed data-labeling and quality-control pipelines for multiple types of data.- Prepared dashboards (Kibana, Tensorboard, D3.js) to explain the results to non-technical staff.- Specified the needs with the business managers and designed company-specific solutions with the product managers.𝐍𝐚𝐭𝐮𝐫𝐚𝐥 𝐥𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐩𝐫𝐨𝐜𝐞𝐬𝐬𝐢𝐧𝐠 (𝐍𝐋𝐏) - Vectorized text using multiple algorithms (TFIDF, Word2Vec, GLOVE, BERT) in order to cluster resumés, detect trends and classify documents. - Developed and trained NLP algorithms (classification, sentiment analysis, opinion analysis, text-matching, clustering, part-of-speech tagging) for purposes such as market research and quality assessment.- Used transfer-learning to train a resumé synthesizer on a small corpus and improve its precision.𝐒𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞𝐝 𝐝𝐚𝐭𝐚- Created a sales-prediction tool using customer sales history.- Improved the sales strategy of a call center using call meta-data and transcripts.- Segmented a customer base using their professional track record to improve email campaigns.

Paris, Île-De-France, France

- Directed the continuous improvement of a receipt-processing API that extracts essential information from images. I reduced response time and memory footprint by 75% and improved accuracy.- Developed deep-learning computer vision algorithms for document processing in Tensorflow (OCR, segmentation, classification).- Designed synthetic data generators to train these models without the need of manually labeled data.- Created a cleaning tool that improved data quality automatically.

Paris Area, France

- Managed a team of two data scientists for work on fraud detection. With financial analysts, we tested supervised (XgBoost) and unsupervised (auto-encoders) algorithms and achieved a recall of 85%.- Developed natural language processing algorithms (sentiment analysis, entity extraction, intent classification) to improve conversational frameworks (Rasa, Watson).- Aggregated and cleaned online posts from various sources (Twitter, Facebook, AppStores, blogs) to prepare training corpora adapted to the mobile banking industry.- Designed a social media post analysis tool for the marketing team.

Paris Area, France

- Developed a natural language processing platform in French and English (Python/Scala).- Built an entity and intents extractor to populate chatbot conversations automatically and reduce the bot design time.- Installed and optimized a parallel calculus framework (Spark) to achieve scalability of the NLP tools.

New York, États-Unis

- Developed a fraud-detection system using machine learning.- Completed IT general-control audits: security reviewing, risk assessment and automation of these processes.- Performed consulting technology missions such as data mining and penetration testing in the energy and financial sectors.

Région De Paris, France

Implemented software solution through customized package configuration (SQL)Developed personalized MI service board to help client assess service performance Designed and Implemented Linux background procedure to ensure system stability and efficiency of client’s retail system (Bash/ Shell)Designed and presented software training to client’s technical team

Atlanta, Géorgie, États-Unis

- Recipient of a fellowship funded by the National Science Foundation.- Elaborated a testing strategy to compare algorithm efficiency in noise-reduction for eye-tracking and head-tracking machines (Matlab).- Designed a machine-learning algorithm to adapt the system to the user's eyes automatically, while reducing noise and response time.

San Diego, Californie, États-Unis

- Developed a low-pressure fluid modeling system used for spacecraft design in a research project sponsored by the United States Air Force.- Refined equation coupling in hybrid multi-physics code using Monte-Carlo simulation (Fortran).- Evaluated and improved the model by comparison with theoretical results.

Master Of Science (Msc), Mechanical Engineering, Gpa: 3.8/4.0

Machine learning, numerical methods, simulation, stochastic processes, data analysis, mechanical engineering

Master'S Degree, Mechanical And Industrial

Design and control of industrial systems, fluid mechanics, electronics, vibration mechanics, statistics

Scientific Preparatory School For Engineering, Theoretical Physics, Algebra, Calculus, Engineering, Chemistry

Academic preparation to the "Grandes ecoles" competitive exam. Major in theoretical physics, mathematics, and engineering.