Robert Dazi

– Developed a Python library to understand how humans look at their environment (eye-tracking) – This project is available on GitHub: https://github.com/v01dxyz/vision-toolkit – Users: scientists, for notebook usage as well as production environments – Development: Refactored a PhD student's code and industrialized it as much as possible – Set up and improve library architecture, interface ergonomics, and ease of use for new users. – Significant QA work to have robust GitHub Actions pipelines, ensuring identical behavior with the old version and the quality of the rewrite. – Preference given to integration tests over unit tests, testing with real data over only fixture-based. – Github Actions pipeline optimization (Docker to minimize environment setup time, ccache to speed up C or C++ compilation, etc) – Libraries used: Numpy, Scipy, Sklearn, Cython – QA/CI: pytest, Coverage, GitHub Actions, Docker/Podman – Read papers for implementing research and algorithms in Python

– Backend Dev: Flask, SQLAlchemy, Celery, Postgres – Data processing: tensorflow, Parquet, Pyarrow, Protobuf, SQL – QA/CI: pytest, Gitlab – Deployment: Kubernetes, docker – Sarus develops a platform providing access to sensitive datasets using differential privacy mechanisms to allow data sharing without compromising personal information. Target customers could be hospitals, healthcare companies, banks or data brokers.

– Data Wrangling using Pandas – Modeling data, improving table schemas to enhance stored data – ETL pipelines with a Postgres DB as processing engine. Maintaining SQL queries for aggregation. – Django to provide a REST API endpoints – Misc.: Docker, Airflow for batch processing, Gitlab CI CD for QA and deployment – CITIO is a startup developping a dataviz tool to help transport operators/transport agencies have a clear idea of how their networks is operated and how their customers use it, providing understable and actionable KPIs (for instance, Monday at 9:00 on a given bus line, there is too few bus for a service considerably impacting the customers feelings) and fine-grained metrics (such as which vehicle is late at a given scheduled trip).