I'm a multilingual AI R&D Engineer & Computer Scientist based in Japan 🇯🇵, passionate about building AI systems end-to-end — from research to real-world deployment. Currently, I am focused on SOTA Computer Vision and industrial AI applications.

• 🔬 World Models - Deepening research into V-JEPA and self-supervised learning to interpret how models build internal representations of complex environments.
• 🧠 E2E Reinforcement Learning & Simulation – Training autonomous control agents using NVIDIA Isaac Lab (PPO / RSL-RL) and differential-drive kinematics.
• 📝 Affective Computing Research – Currently contributing to a comprehensive survey paper on Multimodal Affective Computing.

• 🔬 R&D Computer Vision & Industrial Precision – Design high-precision CV architectures for semiconductor defect inspection, high-resolution microscopic imagery, and edge hardware deployment. Specialized in Semantic Segmentation.
• 🤖 Embodied AI & World Models – Build physical simulation and RL environments in NVIDIA Isaac Lab (PPO via RSL-RL) and explore latent-space dynamics in vision world models (V-JEPA).
• 🧠 GenAI & Agentic Workflows – Implement custom LLM/VLM Agents (Vertex AI ADK), fine-tune models with LoRA, and leverage agentic workflows (Claude Code) for cloud infrastructure tooling and automated media analysis.
• 🛰️ Geospatial AI & Satellite Imagery – Process high-resolution drone and satellite datasets for multi-class segmentation using PyTorch, GDAL, Rasterio, and QGIS.
• 🛠️ Research to Production – Bridge paper implementations (Transformers, ViTs, Swin) with production-grade backends (FastAPI, Flask, PyTest) and interactive dashboards (Streamlit, React).

• V-JEPA Visualizer Dashboard | World Models & Perception
  Open-source unsupervised visual representation visualizer for V-JEPA trained on driving video data to inspect internal predictive latent spaces.
• End-to-End Robotics in Isaac Lab | Embodied AI & RL
  Simulated differential-drive mobile robot navigation using deep reinforcement learning (PPO) in NVIDIA Isaac Lab.
• Swin Transformer & ViT
  Implemented and fine-tuned advanced transformer architectures for image classification tasks.
• Semantic Segmentation with SegFormer & DeepLabV3
  Fine-tuned models for aerial imagery classification (F1 > 90%) using PyTorch.
• Industrial IoT Pipelines Designed end-to-end data monitoring systems using Prometheus, Grafana, and Flask.

Languages: Python, JavaScript, HTML/CSS, SQL, C, Java, PHP
Frameworks: PyTorch, Flask, React, Hugging Face
Tools: Git, Prometheus, Grafana, QGIS, QEMU, Docker
Concepts: Computer Vision, Deep Learning, Transformers, Semantic Segmentation, LLMs, NLP, IoT

🇯🇵 Japanese Business - N2 | 🇬🇧 English Fluent (TOEIC 970) | 🇪🇸 Spanish Native | 🇩🇪 German Basic

• 💼 LinkedIn: linkedin.com/in/alejandro-meza-tudela