Samar Neji

Main Project: SOLIFE (Insurance Platform) Contributed to the development of SOLIFE, an insurance platform, ensuring high performance with large data volumes and complex business logic. • Built and maintained scalable RESTful APIs in Java to support new product features, ensuring efficient data handling and reduced high latency through optimized API architecture and well-structured endpoints. • Collaborated with cloud engineers to support CI/CD pipelines and ensure successful Docker image builds and deployments, gaining hands-on experience with cloud infrastructure and DevOps tooling. • Optimized SQL queries to enhance execution performance and minimize latency, while implementing data archiving and partitioning strategies that reduced report generation time by 35%. • Applied event-driven principles to improve communication between services, preparing the system for asynchronous processing and scalability. • Improved software quality by writing unit and integration tests (JUnit), increasing test coverage from 35% to 70% (as reported by SonarQube). • Led code reviews ensuring readability, Java best practices, test coverage, and performance to ensure high-quality code; maintained Confluence documentation to support onboarding and standardize processes. • Implemented secure authentication and fine-grained role-based authorization, restricting API access based on user roles and permissions to ensure data confidentiality and system integrity. • Served as the primary point of contact for a major feature, coordinating efforts across development, QA, DevOps, and product teams to ensure smooth delivery and alignment. • Mentored and onboarded junior developers, conducting hands-on sessions, collaborative code reviews, and architectural walkthroughs to foster a culture of technical excellence and knowledge sharing. • Ensured efficient front-to-back communication, defining clear data contracts and optimizing request/response handling for better performance.

Project – SOLIFE HELPER (Intelligent Assistant Platform) Developed an intelligent internal assistant to help developers access technical and business information 47% faster. • Automated data collection by scraping internal resources such as JIRA tickets, Wiki documentation, and SQL repositories. • Built data preparation pipelines to clean and structure raw data for model training, integrating Doccano to efficiently annotate and convert this data into labeled JSON datasets stored in a NoSQL database. • Implemented three independent Flask-based microservices, each responsible for training and inference of pre-trained models dedicated to specific tasks: document search, question answering, and text-to-SQL translation. • The backend analyzes each user query to determine its type and routes it to the corresponding microservice for processing. • Developed client and administrator Angular interfaces, communicating with backend services via RESTful APIs for request handling and system management. • Boosted system performance by over 50% by implementing asynchronous task execution for long-running ML operations, leveraging caching (Redis) to store frequent responses, and optimizing document search through indexing techniques. • Used Keycloak for secure user authentication and applied RBAC policies with JWT to control who can access what — only admins can use the admin interface, while regular users can only use the chat. • Deployed the solution using Docker containers and GitLab CI/CD pipelines, ensuring modular scalability and easy maintenance.