Field Robotics · Applied Machine Learning · LLMs · AI-Generated Content
I build systems at the intersection of robotics, computer vision, and modern AI — from multispectral crop/weed segmentation in the field, to 3D retrieval with CLIP + FAISS, to diffusion-based game content and imitation learning for Mobile ALOHA. I enjoy turning research prototypes into things that actually run.
Auto-refreshed weekly from Google Scholar via SerpAPI.
Dataset/notebook counts auto-refreshed weekly from the Kaggle API.
Kaggle is one of my favourite playgrounds for exploring new ML ideas on real-world data. Full profile → kaggle.com/enddl22.
| Project | Description |
|---|---|
| 3D object retrieval tutorial | Toy example for 3D model retrieval from text or image using CLIP embeddings + FAISS indexing. |
| GenAI game scene generation | Unity + 3D retrieval + diffusion + monocular depth estimation demo (playable sample). |
| deepNIR | Synthetic NIR image generation + improved fruit detection with deep learning. |
| Mobile ALOHA (fork) | Imitation learning + co-training for Mobile ALOHA — bug fixes, velocity feature, extra tooling. |
| weedNet | Multispectral crop & weed dataset for semantic segmentation. |
| deepFruits | Kaggle-hosted 11-fruits bounding-box annotation dataset. |
| gym_rotor | Foundational continuous-control RL examples for quadrotors. |
