Tarek Elouaret

- Development of an AI emulation model of a Digital Twin for sequential time series signals, combining LSTM/GRU and Transformer / LLM architectures (representation, prediction, generation/synthesis).

- Setup and industrialization on Google Cloud Platform (GCP): data/compute pipeline management, inference training and deployment, observability and MLOps best practices (versioning, reproducibility, monitoring).

- Improvement of accuracy and robustness on complex signals: processing large volumes, validation strategy, generalization tests, model hardening.

- Agile methodology work: rapid iterations, business/technical alignment, cross-team collaboration.

- Design and development of an accelerated signal processing pipeline integrating AI/ML components (pre-processing + inference) for real-time constraints.

- Implementation and optimization on Xilinx Alveo A80 FPGA: parallelization/pipelining, timing constraints, resource optimization (DSP/BRAM), latency and throughput optimization.

- High-speed network integration: Ethernet / Transceivers, system performance optimization (throughput, response time).

- Missions for international clients via Outlier, Mercor, Upwork, and Turing: application development, rapid prototyping, and AI solution integration.

- Drafting collaboration proposals and technical scoping (objectives, scope, architecture, effort and risk estimation, deliverables).

- Evaluation and assessment of LLM & ML models: quality, robustness, security/alignment testing, error analysis, improvement recommendations (prompts, data, metrics).

PhD in AI — industrialization and performance optimization of an autonomous navigation application through FPGA acceleration and scalable cloud deployment.