Preparing your workspace
Every product here is a live system built and operated by the same team that takes on client engagements. These are engineering case studies, not mockups.
Adaptive learning platform
Multi-tenant SaaS with 94 pages, 600+ adaptive lessons, and 49 widget types. Per-tenant row-level security scopes every database operation. JAX, the resident AI tutor, runs UCB1 routing across 6 model providers with BKT per quiz question and a 4-module H2E adaptive layer (SROI, NEZ, IGZ, V-RIM). Backed by 27 production Supabase edge functions and a governed course builder with durable job queues.
944 commits · 27 edge functions · 49 widget types
Event photo/video + media platform
61+ screens in Expo React Native, live on the App Store and Google Play. AI moderation pipeline, real-time event media collection, and four white-label client instances deployed at separate domains. The mobile build demonstrates the same multi-tenant, governed architecture carried into a native runtime.
561 commits · App Store + Google Play · 4 white-label instances
AI video commercial platform
Text prompt to broadcast-ready commercial in a single pipeline: Vertex AI Veo for video generation, Shotstack for programmatic assembly, ElevenLabs for voiceover synthesis. Real-time generation tracking surfaces progress at each stage. Designed for commercial production cycles where manual editing is the bottleneck.
317 commits · Vertex AI Veo · Shotstack · ElevenLabs
Voice AI business automation
LiveKit full-duplex transport paired with Gemini Live inference and Google Calendar scheduling. SIP telephony handles inbound calls; the system qualifies, schedules, and warm-hands off to a human when escalation thresholds are met. Admin dashboard surfaces session logs and escalation history.
LiveKit · Gemini Live · SIP telephony · warm-handoff escalation
Local-first personal AI operating system
On-device knowledge graph, retrieval layer, and ML pipeline running entirely on Apple Silicon. 61 Python modules covering semantic memory, symbolic memory, relational memory, language modeling, and behavior intelligence. Zero cloud dependencies for inference — private by construction.
61 modules · on-device inference · Apple Silicon · 100% private
Custom build
The same engineering team that built and operates every product above takes on client engagements. Production software, AI and ML systems, digital media pipelines, and training platforms — scoped around the actual operation, not a generic template.
Start a project → /start
The same team that shipped these systems is available for client engagements — software, AI, media, and training platforms built to operate.