Data & AI Engineer , Capgemini Engineering
Data & AI Engineer
Capgemini Engineering
2021-03-01 – •
● Document Intelligence for Aircraft Manuals (Client: Airbus):
– Designed and prototyped OCR-based python pipelines to detect pictograms and numeric data in aircraft manuals.
– Cut manual review time by 40% and improved early error detection.
– Optimized workflows for CPU-only systems to enable on-site deployment.
– Built a Streamlit interface to visualize results and deliver interactive demos to clients.
● High-Performance Computing Activity Monitoring (Client: Airbus Helicopters):
– Designed and deployed an automated dashboard, drastically reducing manual work and improving operational visibility.
– Delivered faster insights through clear, user-friendly data visualizations.
● Transformer Model Optimization for CAD workloads (Capgemini Engineering R&D):
– Researched optimization techniques for Transformer models used in CAD (e.g. DeepCAD), targeting CPU-only environments.
– Applied quantization and pruning via Intel Neural Compressor, achieving 50% size reduction and 92% faster inference post fine-tuning on AWS.
– Wrote internal documentation to support future R&D efforts on lightweight AI models.
● Reinforcement Learning for Aircraft Trajectories (Client: Airbus):
– Improved trajectory performance by 10% through fine-tuning of reward functions and hyperparameters.
– Documented and transferred knowledge effectively, enabling the client to build upon the solution independently.
● Energy Efficiency Analysis for Inference Workloads (Capgemini Engineering R&D):
– Analyzed how hardware choices (CPU architecture, clock speed, model FLOPs) influence energy efficiency, helping guide cost-effective deployment strategies.
● Human Emotion Recognition Analysis (Internship):
– Developed a multimodal AI system capable of recognizing human emotions using text, audio, and video inputs.
– Gained hands-on experience with cross-modal feature engineering and deep learning models.

