AI SWE @ Arista Networks | AI/ML @ EPITA | Ex-Cisco | AWS/Azure Certified
Current work: Agents & world models + state space architectures (with Dreamer4-Mamba-JEPA )
On-hold: A2A communication, benchmarking RAG methods, SNNs for edge-neural networks, and lots more I need to get back to once the above is finished!
Shoot me a message if any of this sounds interesting, I'd love to talk more :)
| Project | Description | Tech |
|---|---|---|
| Dreamer4-Mamba-JEPA | Training agents inside of scalable world models. | Python, CUDA |
| ThreeBody-JEPA | Benchmarking world models (JEPA) on their extrapolation of chaotic three-body dynamics. | Python, PyTorch |
| Mamba Reimplementation | From-scratch reimplementation of the Mamba paper, exploring state-space model efficiency. | Python, PyTorch |
| Encoder-Decoder Translator | Bidirectional GRU Encoder-Decoder translator built from scratch in CuPy. | Python, CuPy |
| KD4Exo | Predicts exoplanet habitability using knowledge distillation (99.98% accuracy, 99% reduction in training time). | Python, CUDA |
| Tiger Compiler | Full compiler pipeline (lexer, parser, semantic analysis, LLVM backend). | C++, LLVM |
| POSIX Shell | Robust shell implementation with process control, I/O redirection, and memory safety. | C, Valgrind |
Languages: Python, C, C++, C#, Java, JavaScript, Assembly
AI & Data: PyTorch, TensorFlow, LangChain, Scikit-Learn, Hugging Face, CUDA
Cloud & DevOps: AWS, Azure, Docker, CI/CD, Git, Linux
Certifications: AWS Cloud Practitioner • AWS Solutions Architect • Azure Fundamentals • Azure Administrator
