AIR Lab (Aviation Innovation Research Lab): Advancing Global Aviation efficiency and sustainability by creating innovative Air Traffic Management solutions through open technologies and collaboration.- Developed a component of a data ingestion product enabling internal users to store data in the Datalake. Users declare the data source through a GitLab pipeline, that verifies validity and eventually posts the schema to Apicurio, allowing for data cataloging and automating the delivery process. Tasks involved: . Releasing Apicurio Schema Registry instances in AKS clusters using Helm. . Writing Helm chart and release for a Configmap Watcher Cronjob. . Deploying ingresses to enable the connection between Apicurio endpoints and Gitlab CI. . Writing the Gitlab CI jobs: validate, convert, post. . Implementing Keycloak and OAuth2 protocol for user authentication and authorization management.- Developed a CLI (Command Line Interface) for Azure Data Lake Storage in Python to speed up operations between the data team and ADLS.

- Developed an advanced chatbot application tailored for multiple use cases. It is aimed at assisting consultants in their daily operations. It leverages private company data and the latest OpenAI models to generate unique and precise responses. Tasks included: . Creating python pipelines to collect data (.pptx, .docx) from Sharepoint, clean it, format it and store it in Azure Blob Storage. . Extracting all text elements from raw files (including in tables, plots and graphs) store into Azure Vector Store Database. . Developing the app in python with features like rating the response, choosing the LLM, opening the document mentioned in response, choosing the response style and more... - App deployment, data, and models are managed through different Azure services (Azure OpenAI, Azure Blob Storage, Azure App Service).- Showcased during the organization's yearly seminar (+1500 consultants). Led to new business opportunities.

-Developed two NLP solutions and a web application with the goal of translating natural language questions into SQL queries using python. The first one is using a fine-tuned version of the T5 language model. The other is using the GPT3 model and the few-shot learning approach.-Developed an API displaying data to customers with multiple filters capabilities using python and deployed with Azure Functions in order to facilitate data exchanges between clients and the company.

- Built crypto trading bots, using Machine Learning models.- Learned trading and technical analysis theory.- Practiced trading on a TradingView paper account.