Building for
human and AI
collaboration.
I ship agent-based solutions 0->1 that embed seamlessly into enterprise and developer workflows, taking a customer-focused approach to build products from discovery to production. Currently building Athena AI at Sony.
Four layers, one workflow.
Each layer builds on the platform beneath it — from the collaborative planning surface where teams align, down through the no-code experimentation sandbox, the workflow integrations that move plans into code, and the core platform tooling that powers it all.
Collaborative Planning
A shared planning layer for turning fragmented information into diagrams, artifacts, and workflow decisions teams can actually reuse.
No-code Experimentation
A guided experimentation sandbox for testing, prototyping, and debugging workflows with AI assistance.
Workflow Integrations
Local tooling and workflow handoff products that move planning cleanly into real implementation environments.
Platform Tooling
Core APIs, retrieval infrastructure, and agent tooling — the foundation layer that makes the rest of Athena usable in real workflows.
Agentic Architecture for Athena
How I designed Athena as a product-facing agent system: workflow-first orchestration, explicit artifacts, and UX that helps humans and AI collaborate.
Building RAG Systems That Are Actually Usable
A practical look at the retrieval layer behind Athena: ingestion, metadata, reranking, MCP tool surfaces, and why good RAG systems are workflow products.