Frederic Ferriere

Feasibility study of a urban logistics project whose idea was to combine public transports and bike delivery to facilitate parcel delivery at large scale without the need for LUV. * Full back & front end development, using a mix of algorithms and AI for multimodal synchronisation * Technical and regulatory feasibility study * Meetings with different actors & attending salons to gauge market expectations Stack: Python, OSRM, Streamlit, Pydeck The project was dropped due to organisational and regulatory hurdles that could not be overcome in a near future.

mySofie is a mobile app helping users have a better understanding of their health spending. By aggregating financial flows from Social Security with those of your private insurance, it allows a simple representation of the true cost of health insurance. As the first data recruit, my role was to identify useful data sources, make them more accessible and interact with the business to identify potential use cases. It required a mix of data science, data engineering and data analytics. Achievements: * 50% cost decrease for user acquisition in a high growth environment (from 50k users to 300k users), using clustering techniques for user segmentation * +3 points conversion rate in the onboarding process following a thorough analysis of the conversion funnel crossing data from various sources (Mixpanel, internal PostgreSQL db, Google Analytics) * Easier access to data following the setup of a BigQuery db combiening Google analytics tables and external tables fed via Google Cloud Storage * Automation of data ingestion from various sources (Mixpanel, Google playStore) using Powershell scripts and local jobs * With a backend engineer, cocreated a self service tool to automate data extractions from our postgreSQL db using parameterised jobs (parameters being frequency + query string), with results saved to Google Cloud Storage * Created a product analytics dashboard leading to weekly analysis of the performance of different acquisition channels

Worked on a carpooling solution aimed at identifying the number of passengers in a vehicle using computer vision and artificial intelligence. Tech stack: OpenCV, Keras, RabbitMQ, Linux, Python, C Skills gained: Git, Agile, computer vision, deep neural networks