I'm James Keane.

Built production AI systems and agentic workflows. YC W23.

A multi-agent editorial pipeline that transforms raw event streams into structured intelligence analysis.

A real-time geopolitical event monitoring dashboard with a 3D interactive globe and AI analysis.

An open-source machine learning framework and transpiler for PyTorch, TensorFlow, JAX, and NumPy.

A 160km Himalayan trek culminating in the crossing of the Thorong La pass at 5,416m.

A highly technical alpine ridge traverse summiting Mount Okuhotaka (3,190m) and the Gendarme.

A 170km alpine circuit featuring over 10,000m of cumulative elevation gain.

A multi-day island crossing navigating steep volcanic ridge lines and dense alpine forests.

A continuous 35km single-day endurance route summiting 15 peaks across the range.

A 120-meter WWII shipwreck dive featuring structural swim-throughs and dense marine life.

High-current drift dives navigating underwater pinnacles and pelagic drift zones.

Open ocean cleaning station dives focusing on large pelagic species encounters.

Advanced deep-water and navigational training across multiple reef sites.

Upper Second Class Honours (2.1).

Foundational principles of distributed systems, database architecture, and data pipelines.

An iterative framework for framing, building, and deploying reliable ML models in production.

Architectural patterns and best practices for building robust applications with foundation models.

A practical guide to customer extraction, avoiding bad data, and identifying true user intent.