Building an AI-powered compliance platform that takes companies from zero compliance infrastructure to SOC 2 certification in weeks, not months.

The outcome: Customers consistently achieve SOC 2 Type I in 3-4 weeks with zero audit exceptions - versus the industry standard of 3-6 months. Onboarding takes under 30 minutes. Total customer effort is typically under 5 hours.

What I built:

• Continuous monitoring system - Real-time infrastructure compliance via OIDC integrations with AWS, GCP, and Azure. Alerts when controls drift (MFA disabled, encryption gaps) and automatically collects evidence proving compliance.
• AI control generation - Upload organizational documents, get framework-mapped controls. The entity mapping pipeline can onboard a new compliance framework (including custom client frameworks) in hours rather than manually-reviewed months.
• Cross-framework mapping - Built the canonical control system that maps between frameworks such as SOC 2, ISO 27001, HIPAA, GDPR, and NIST CSF. One piece of evidence can satisfy multiple framework requirements automatically.
• Full platform - Next.js frontend, TypeScript API, Python AI microservices, PostgreSQL, deployed on AWS with support for on-premise installations.

Background: EasyAudit acquired Veita (see below) in early 2026 - we'd been their primary engineering partner, and the acquisition brought the product fully in-house.

Lead engineer helping enterprises transform their operations with AI co-pilots and automated workflows. Built the core platform featuring context-aware AI agents and a visual workflow editor, customized for specific business processes, data sources, and team requirements - including HR chatbots for Canadian hospitals. Working closely with consultants, I took solutions from proof-of-concept through to robust production systems.

Acquired by EasyAudit, our primary client, in early 2026.

Six years building infrastructure for autonomous digital humans - AI-driven avatars that see, hear, and respond in real-time.

Digital Human Creation Pipeline

The outcome: Reduced digital human creation time from 6 months by a ~20-person team to 30 minutes in-browser by an end user.

How:

• Automated artist processes - Took manual, specialist workflows and made them algorithmic
• Built the "baking" microservice - Finalizes user creations in the Avatar Designer web app, turning configured assets into deployable digital humans
• Internal + production infrastructure - Systems for creating and deploying 3D assets and Digital Brain models
• Unity migration - Rebuilt the entire 3D asset pipeline for Unity, replacing a proprietary OpenGL renderer. The old system required specialized artists trained on custom tooling; Unity opened standard pipelines and self-service creation.

Conversational Architecture

A digital human is an AI agent on a real-time video call and to help users, it needs to orchestrate other agents in the background: Salesforce, ServiceNow, OpenAI etc.

Say your flight just got cancelled. You need to rebook, find a hotel for the night, and notify your meeting that you'll be late - three tasks, three backend agents, all running in parallel. But voice is sequential; unlike a text-based chatbot, you can't show a wall of status updates. The digital human has to deliver information empathetically, one thing at a time, while tracking what's been said, what's pending, what the user actually knows and what thread/task the user is currently talking to.

To solve this, I designed and built a system that coordinates parallel backend agents with sequential voice delivery.

• Single source of truth - Agents, UI, and conversation all read/write to the same state
• Synchronized delivery - UI appears exactly when the avatar speaks about it
• Graceful interrupts - User changes priorities mid-conversation, in-flight work is superseded
• Multi-task orchestration - Backend agents collaborate without overwhelming the conversation