Ayoub E.

- Developed and maintained an automated anomaly detection system for time series data sourced from nuclear power plant sensors, electricity trading markets, and other energy sector systems, handling both the application and infrastructure.

− Designed and implemented ETL data pipelines using Python, Kafka, Polars, and Parquet to retrieve and process time series data from various sources.

− Optimized the anomaly detection scoring algorithm by developing C extensions with Cython, enabling GIL removal to fully leverage parallelization with OpenMP, achieving a 300x speedup by reducing processing time from 4-5 min to just 1 sec.

− Developed REST APIs using FastAPI to expose anomaly detection algorithms, enabling real-time identification of anomalies across diverse use cases.

− Led a team of three people while communicating with stakeholders, including end users such as traders, teams monitoring nuclear stations, and other operational groups, on progress updates, collecting suggestions, and working to resolve specific software bugs.

Technologies used: Python, Cython, Kubernetes, OpenShift, Kafka, GitLab CI/CD, FastAPI, Pydantic, Asyncio, Polars, Parquet

- Built and maintained QuarksFlow an automated file and URL analysis platform for malware detection.

- Migrated the malware detection scan system from a DAG workflow-based architecture to an event-driven system leveraging shared probes microservices, resulting in a 40% reduction in scan execution time and improved CPU and memory resource usage.

- Integrated a custom Kubernetes event-driven autoscaler for the shared probes microservices, enabling dynamic scaling in response to workload fluctuations. This optimization helped optimize resource allocation and ensure efficient utilization of infrastructure resources

- Designed and implemented a Kubernetes controller to manage probes resources and their associated Custom Resource Definition (CRD) scaler objects.

- Continuous Integration/Deployment Pipeline Integration, Code Reviews, Load Testing, Unit/Integration/E2E Testing.

Technologies used: Python, Go, Docker, Kubernetes, Helm, minikube, Redis, PostgreSQL, MinIO, Keda, Argo, Rancher, Azure, GitLab CI/CD

Automated Game Testing for bugs detection through Novelty Search and Quality Diversity algorithms (Application to Space Engineers game).

- Implemented a 3D maze game environment similar to Space Engineers game with a navigation character using the physics engine SDK Bullet.

- Implemented the API layer between the 3D game environment and the learning algorithms using OpenAI Gym.

- Developed and experimented with different Novelty Search algorithms and their variants on the environment to benchmark the coverage/exploration of the behavioral space, in order to automate bugs detection in the game environment.

Technologies used: C++, Python, Gym, Pybullet, Tensorflow, Keras, Git