Hi, I'm Walid Abouir.
A
Self-driven, passionate about data science and Artificial Intelligence with a curious mind who enjoys solving complex and challenging problems.
About
I am an Applied Mathematics Student at INSA Toulouse. I enjoy problem-solving and coding. Always strive to bring 100% to the work I do. I have worked on technologies like Python, C++ and R, during my studies, and also during the various projects that I had to accomplish. I am passionate about exploring data by using Artificial Intellingence fields that solve real-world problems impacting different users.
- Languages: Python, R, C++,
- Libraries: PyTorch, Tensorflow, Gymnasium, Stable-Baselines3, Pandas, Scikit-learn, Matplotlib
- Tools & Technologies: Git, Linux, LaTex
Looking for an opportunity to work in a challenging position combining my skills in Data Science and Machine Learning Engineering, which provides professional development, interesting experiences and personal growth.
Experience
- Performed a technical and scientific literature review on Safe RL, Constrained RL.
- Studied the operation of microgrids, independent power networks supplying electricity to a residential building at École Polytechnique (Bachelor)
- Trained LSTM and Attention-based models (Transformers) for features extraction from the Bachelor environment data, used to train a Soft Actor Critic (SAC) agent to reduce daily electricity bills.
- Extended an existing microgrid environment to train RL agents under operational constraints, a key step for real-world deployment.
- Reimplementing the Lagrangian-SAC algorithm from literature to train an agent on a CMDP version of a previous environment.
- Implemented Shielding techniques and Control Barrier Functions to ensure the agent’s safety.
- Reduced the rate of operational constraint violations by 80% using Safe RL agents (Lagragian-SAC).
- Tools: Python, PyTroch, Linux, LaTex, Wandb, Gymnasium.
- Developped a Deep Reinforcement Learning Gym environment for the ACAS-Xu.
- State of art on RL, Deep RL and Safe RL algorithms.
- Model training on different environments or toy cases to familiarize myself with RL and Deep RL algorithms (Cartpole, FrozenLake, Pyflyt).
- Creation of the Gymnasium ACAS-Xu environment for collision avoidance between unidentified drones, opening up new avenues of research, for example in Multi Agents for ACAS-Xu, and reducing the collision percentage by 85%.
- Tools: Python, Linux, LaTex.
- Worked on stochastic optimization problems using Python.
- Read and extracted theorical informations form scientific publications.
- Coded and solved the salesman stochastic optimization problem on Python.
- Tools: Python.
Projects
Reasearch & innovation : Calculating the magnetopause with Symbolic Regression.
- Tools: Python, R
- Read scientific articles about physics plasma and Symbolic regression.
- Explore data sets provided by the Onera laboratory and use Pandas, Scikit-learn libraries.
- Use of the PySR library to caculate Schue's magnetopause model.
Data Science project on the Vélib network of Paris.
- Tools: Python, Github, R
- Exploring the velib data set in Python : Statistical analysis (PCA), clustering data.
- Doing the same step in R.
- Coordination of the group work on Github
Skills
Languages and Databases
Python
R Language
Java
Ada
C
Shell Scripting
Libraries
PyTorch
Stable-Baselines3
NumPy
Pandas
Seaborn
Scikit-learn
Matplotlib
Other
Git
JIRA
Linux
Education
Institut National des Sciences Appliquées de Toulouse
Toulouse, France
Degree: Master of Science in Applied Mathematics
Dates: July 2020 - Present
- AI Frameworks
- High Dimensional Deep Learning
- Statistics
- Machine Learning
- Gaussian Processes
- Optimization
- Data analysis
- PDEs and Mechanic of fluids
Relevant Courseworks: