I'm a computer vision researcher finishing my MASc at Queen's University, where my work focuses on 6D object pose estimation — teaching machines to understand not just what objects are, but exactly where they are and how they're oriented in 3D space.
I build end-to-end systems: from training custom segmentation and regression models, to classical geometry (RANSAC, ICP, Horn's method), to deploying inference pipelines in real-time robotic environments.
Lately I've been extending into multimodal AI — combining pose estimation with LLMs to enable natural language interaction with 3D scenes.
- 🎓 Finishing my MASc thesis on template-based pose estimation (code open-sourced upon publication)
- 🔧 Building a multimodal pipeline: pose estimation → structured scene representation → LLM reasoning
Core: Python · PyTorch · OpenCV · Open3D · NumPy
ML/AI: Deep Learning · Instance Segmentation · 6D Pose Estimation · LangChain · RAG · HuggingFace
Tools: ROS2 · Docker · W&B · PyTorch Lightning · SLURM
| Project | What it is |
|---|---|
| bin-picking-pose-estimation | 6D pose estimation for robotic bin-picking — IBPC 2025, OpenCV sponsored |
| LawQA-RAG | Legal question answering over 100k documents using Llama-2, LangChain, Pinecone |
| Blenderproc-Data-Generation | Synthetic RGB-D data generation for pose estimation training and template building |
📍 Kingston, Ontario — open to ML/AI engineering roles
