MINDSIGHT : The Development and Deployment Of A Machine Learning Based Mental Health Prediction Platform

As a Software and Machine Learning Engineer at Medico Call, I led a 5-members team, I played a pivotal role in designing the architecture of the entire project, including the database, to facilitate the development of a machine learning-based model for predicting pathologies after processing and analyzing data and psychological tests. I actively contributed to the design and implementation of the platform's architecture, ensuring seamless integration between different components.

• Led architecture design, ensuring scalability, efficiency and robustness system-wide, enabling application to be scaled up to be used across population of 10,000+ users.
• Developed RESTful API using Python Flask, using SQLAlchemy and Alembic for integration and migration of underlying SQL database, streamlining management of user credentials and test information.
• Conducted extensive exploratory data analysis on complex psychological datasets, uncovering critical patterns.
• Built a machine learning model that allows processing and analysis of psychological tests to predict pathologies, achieving a 92% accuracy rate.
• Designed and implemented a RESTful API using Python Flask delivering ML solutions, seamlessly integrating SQLAlchemy and Alembic for agile database management and migration.
• Designed and implemented a data visualization dashboard using PowerBI, enabling stakeholders to monitor key performance metrics in real-time.
• Collaborated with data scientists to implement XGBoost algorithms, reducing prediction errors by 25%.
• Optimized database model using SQLAlchemy ORM and Alembic for database migration, ensuring data integrity and consistency.
• Introduced JWT-based authentication and authorization measures, securing API to provide access only to authorized users.
• Used machine learning algorithms including logistics regression, decision trees, gradient boosting and random forests to analyze user data and predict pathologies.
• Employed recurrent neural networks for advanced feature extraction and identification of new pathologies and anomalies.
• Developed Python script that ran continuously to analyze user data and send daily recommendations to users with mental health issues based on predicted pathologies.
• Deployed API using Heroku; ensured seamless integration and deployment by connecting API to GitHub repository, allowing for hassle-free updates and enhancements.
• Utilized Scrum methodology and employed Atlassian Jira, Confluence, and Bitbucket for project management, sprint planning, issue tracking, and code management.

Technologies Used: Python, Flask, MySQL, SQLAlchemy, Alembic, Git, Heroku, XGBoost, RNNs, Jira, Confluence, Bitbucket, Scrum