Work

Selected systems from client engagements and internal platform development.

Everything shown here is built and operational — not conceptual demos.

A private, containerized assistant that runs locally with document ingestion + vector search.

Client needed a private assistant to chat with internal documents without sending data to external APIs.

Working private assistant with document Q&A and semantic search, delivered with deployment instructions and a reproducible Docker Compose setup.

Client name withheld. Screenshots and technical walkthrough available on request.

Internal platform for designing and deploying personalized AI systems tailored to real workflows, data, and constraints.

Clients need AI systems that reflect how they actually work — not generic chatbots or one-off automations.

Modular composition of specialized AI "brains" with tools, memory, and coordination
Designed for long-running workflows and system-level control
Adapted per client use case without exposing underlying complexity

Ongoing internal platform that enables faster, more reliable delivery of custom AI systems for client engagements.

Internal platform. Not offered as a standalone product.

A web application that applies audio effects to uploaded files using natural-language instructions.

Musicians and creators want to manipulate sound without navigating complex DAWs or effect chains.

Audio upload and processing pipeline
Natural-language interpretation of effect intent
Effect rendering using Pedalboard
API-driven backend coordinating AI interpretation and deterministic audio processing

Working web application that transforms audio based on user prompts, demonstrating end-to-end AI-driven creative tooling.

Let's talk about your system

If you have a real operational problem, we can discuss architecture, feasibility, and implementation.

Initial calls are exploratory and do not constitute a service agreement.