Description
Summit Seeker is a collaborative project with Alessandro Arensberg focused on making mountain exploration more accessible and actionable. We designed a streamlined platform where users can discover the most efficient way to climb a natural rock wall. To achieve this, we fine‑tuned our own YOLO-based model, PRISE, to visually detect holds and suggest optimal ways to use them.
Key Features
- Summit Discovery: Browse and search mountain spots with structured information, difficulty signals, and quick comparators.
- Trip-Oriented UX: Designed flows centered on decision making, from discovery to route planning.
- Data Enrichment: Integrated external data sources to provide richer context around each summit.
- Responsive Interface: Built a layout that remains smooth and readable on mobile and desktop.
Technologies Used
- Python : Language used to develop the entire project
- PyTorch : Training model and create our own fine tune YOLO model
- Albumentation : For data augmentation
- Git: Version control and collaboration with Alessandro Arensberg.
My Role
Summit Seeker was a collaboration with Alessandro Arensberg. I focused on the model training and the data integration patterns that power the summit pages. Our model was a custom YOLO model called PRISE to detect climbing holds from photos.
What I Learned
Building for a niche audience — climbers — forced us to be precise. A generic route planner would not work; we had to understand grades, exposure, and hold types. It reinforced my belief that the best products come from knowing your user, not just your dataset.
Outcome
The project delivered a practical and elegant base product for mountain discovery, with a strong emphasis on clarity, performance, and collaborative product thinking.