Ryte is a Canadian start-up in the healthcare sector, aiming to connect physicians and patients across North America. I worked on various tasks, primarily in NLP, Computer Vision, and MLOps.NLP:- Created a RAG system to generate biographies from internet data, including R&D to accelerate LLM inference and reduce hallucinations.- Developed mapping and disambiguation algorithms to link different data tables (96% of data mapped).Computer Vision:- Implemented algorithms to detect faces in photos collected from the internet and enhance image quality when necessary.- Conducted R&D on image distortion in restored images by our models.MLOps:- Built a pipeline for automating the mapping of medical data.- Set up CI/CD and unit tests, with successful deployment of the pipeline into production.Technologies used: Python, PySpark, SQL, Databricks, Azure DevOps, Git, Python Wheel, PostgreSQL, Docker

Vectors Group is a communication agency focused on analyzing and improving the online reputation of its clients. Within the data team, I worked on NLP and data engineering challenges.NLP:- Sentiment Analysis: Conducted a study of Twitter user reviews on the client over a given period.- Topic Modeling: Generated the main discussion topics on Twitter for the same period to compare with the sentiment analysis.- Developed a RAG system to automate and enhance the sentiment analysis of tweets for the study.Benchmarking:- Evaluated different sentiment analysis models to select the most effective one for our use case.Data Engineering:- Developed a scraping algorithm to automatically collect daily newspapers (national, regional, political, and economic press) to feed the knowledge base of a chatbot for the company.- Imported raw data into a CSV file usable by the chatbot.Technologies used: Python, SQL, Linux, Git, Meltwater.