Toutel NTATLA YONGUET

– Integrated Go-based RAG/LLM services with Python REST APIs, enhancing async job handling and retries. – Developed end-to-end orchestration for AI-assisted PPTX content generation from PDFs, ensuring structured outputs. – Automated PPTX generation using python-pptx, adhering to brand guidelines and improving production readiness.

Project: Backend Development – Predictive Maintenance Platform for Power Plant Equipment Contributed to the backend development of a predictive maintenance application for power plant equipment, working within a microservices architecture in an international, Agile environment. Key contributions: – Took ownership of multiple existing microservices. – Developed new features across several backend services. – Built containerized APIs (FastAPI, Flask, Django with Docker) to interconnect services. – Utilized various AWS services including Lambda, Step Functions, ECS, S3, and DynamoDB. – Wrote unit tests using Pytest to ensure code stability. – Participated in code reviews and pair programming sessions. – Handled CI/CD pipelines and production deployments via GitHub Actions. – Worked in an Agile environment: daily stand-ups, sprint planning, and sprint reviews. – Collaborated in a bilingual (French/English) international team setting.

Project: Backend Refactoring – Monolith to Microservices Architecture Redesigned the backend architecture of an existing application, transitioning from a monolithic structure to a modular, microservices-based system. Key contributions: – Audited the legacy backend code to identify and define microservice boundaries. – Introduced a layered architecture to enhance modularity and code maintainability. – Designed and implemented the foundational structure for multiple microservices. – Managed data storage using a NoSQL database (DynamoDB). – Developed the insurance policy management microservice with FastAPI. – Built various RESTful APIs within the microservice. – Wrote comprehensive unit tests using Pytest to ensure code reliability.

Project: Development of a POC to optimize the charging of electric cars Key contributions: – Training to understand the basics of the field – Write technical and functional documentation – API development (FastAPI) – Data management via a SQL database (PostgreSQL) – Add unit tests for new functionalities (pytest) – Deployment of code on Microsoft Azure (Microsoft Azure functions) – Live testing with customer

Projects: Python & Django Development – High-Performance Computing Tools 1st project: Terminal-based Task Manager for TGCC Supercomputer (Python + Flask) Developed a Python terminal application to replace an outdated Bash script used to manage computation jobs on the TGCC cluster (managed by SLURM). The new tool provided enhanced interactivity and user control. Key contributions: – Built a terminal-based interface allowing authorized users to edit pending jobs in SLURM. – Integrated LDAP authentication for access control. – Connected the frontend to SLURM via a Flask-based API layer. – Implemented an automatic refresh mechanism to display real-time job status updates. 2nd project: Web Application for TGCC User Account Management (Django + Celery) Created a Django-based web application to manage user accounts required for accessing the TGCC supercomputer. Key contributions: – Designed asynchronous user creation workflows using Django and Celery. – Enabled monitoring of user-related KPIs through a web dashboard. – Built admin features for managing and updating user accounts. – Contributed to frontend development by creating and updating HTML templates when needed.

Project: PDFCrawler – Intelligent Document Analysis Tool Designed and developed a Python-based tool named PDFCrawler to support engineers in processing and understanding large volumes of newly received PDF documents related to an infrastructure project. Key objectives addressed by the tool: – Identify critical documents within large, unstructured sets. – Classify documents by domain (e.g., civil works, electrical engineering, or both). – Provide a user-friendly interface to help engineers quickly grasp the content and relevance of a document set. – Handle scanned (non-parsable) documents by implementing fallback strategies. – Extract contextual information from uncodified documents to infer their category or field when metadata is missing.

Project: Predictive Modeling for Hypertension Treatment (Academic Research Traineeship) The goal was to develop a predictive model to determine the most effective hypertension treatment for individual patients based on clinical data. Key contributions: – Implemented supervised learning algorithms to predict optimal treatments based on patient datasets. – Evaluated model performance and identified limitations due to high error rates. – Researched and designed a custom bagging-based ensemble approach to improve prediction accuracy. – Explored the use of a multi-agent system, where each agent produced individual predictions and collaborated with others to adjust and improve the overall outcome. – While time constraints limited the full implementation of the multi-agent approach, the project laid a strong foundation for future work.

Project: Development of Academic Information Systems – AgroParisTech As part of a team at the National Institute of Agronomy, I contributed to the development of a module for the new Digital Work Environment (ENT), which is set to replace existing academic systems across several French institutions, including Télécom ParisTech, ENSTA ParisTech, and École Polytechnique. During the second half of the internship, I collaborated on the complete redesign of the faculty skills directory for AgroParisTech. This team project also served as the subject of my Bachelor's final project.