I like building things that make people's lives a little easier.

Problem

As AI agents start working across teams and systems, dm needs to know who owns them, what they can do, and how their access and actions are reviewed.

Approach

Designed a company-wide architecture connecting agent platforms with existing and future identity services, plus a central policy layer that can allow, block, or escalate sensitive actions.

Outcome

Gives teams one shared foundation for managing agent access, approvals, and audit across platforms, instead of solving the same governance problem for every new use case.

Problem

Internal teams needed a reliable way to build and operate their own generative AI and agent-based applications.

Approach

Led seven engineers in a 13-person team to build a serverless AWS backend and agent gateway, integrating internal platforms across AWS, Azure, and GCP.

Outcome

Gave internal teams a shared foundation for shipping generative AI and agent-based applications.

Problem

Incoming email across global logistics workflows needed to be classified reliably at high volume.

Approach

Built and deployed BERT-based classification services using AWS Lambda, ECS, API Gateway, and DynamoDB.

Outcome

The system processes more than 190,000 emails each month in production. The work received 2nd place in WirtschaftsWoche Best of Consulting 2020, Technology, with Ginkgo.