akeno.ai

akeno.ai (formerly Data Lighthouse) is a supply chain traceability platform: enterprises use it to trace materials and products through complex, multi-party supply chains.

The problem

Supply chain data is sprawling across suppliers, batches, certifications, and schedules, and the people who need answers aren't database experts. The platform needed real-time visibility over that data, interfaces that could handle bulk operations and complex scheduling, and eventually a way to just ask questions of it in plain language.

What I built

Four years of full-stack product work, from the data layer to the AI assistant:

• The AI assistant feature, built on OpenAI via Azure and LangChain, with conversational flows that answer supply chain queries using context from the platform's own data.
• Real-time data plumbing: Hasura GraphQL subscriptions and RESTful endpoints, with API contracts documented and kept consistent across the codebase.
• Heavy-duty interfaces: AG Grid and Bryntum extended well beyond their defaults for complex scheduling views and bulk data management specific to supply chain workflows.
• Product analytics with Amplitude, tracking feature usage that fed directly into decisions about what to build and what to drop.
• Sustained refactoring of key parts of the codebase to clear technical debt as the product grew, plus proof-of-concepts (documented with findings and trade-offs) whenever the team evaluated new tools.

I worked directly with the CEO on feature scoping, translating requirements into technical plans and pushing back when something needed more clarity before building started.

Stack

ReactJS, Next.js, TypeScript, Tailwind CSS, GraphQL, Hasura, Node.js, Python, OpenAI API, LangChain, Microsoft Azure, Cypress, AG Grid, Bryntum, Amplitude.

Impact

The AI assistant turned a specialist tool into something anyone in the business could query. Amplitude-instrumented features gave the product team evidence for roadmap calls, and the ongoing refactoring kept a four-year-old codebase shippable as it scaled.