Applied AI and agents
AI that runs in production.
Agents, retrieval, and the testing that keeps them honest. See AI Agents to Production.
Claude, OpenAI, and Gemini APIs
Open-weight models (Llama, Hermes)
Multi-agent systems
Model Context Protocol (MCP)
Agent2Agent (A2A)
Retrieval-augmented generation (RAG)
Vector search (pgvector, Pinecone, FAISS)
Knowledge graphs (Neo4j)
Evals and regression testing
LLM classification
Semantic search
Prompt engineering
Workflow automation
The busywork, wired together.
Lead intel, inboxes, CRM updates, texts, and the hand-keyed steps between your systems.
n8n
Zapier
Webhooks and REST APIs
Twilio (SMS)
SendGrid (email)
HubSpot
Salesforce
QuickBooks
Firecrawl
Tavily
Apollo.io enrichment
Hardware and physical prototyping
When the answer is a thing you can hold.
A sensor on a machine, a kiosk at an event, a device prototype you can test before you fund it. Arduino and Raspberry Pi on the bench, computer vision at the edge.
Device prototyping (Arduino)
Device prototyping (Raspberry Pi)
Edge AI on NVIDIA Jetson
Sensor integration
Physical computing
Interactive installations
Body tracking and gesture recognition
Green screen and chroma key rigs
Print automation
Device tethering
Thermal imaging analysis
Augmented reality
3D product configurators
Computer vision and documents
Cameras and paperwork, read by machines.
PyTorch
TensorFlow
OpenCV
YOLO
Hugging Face
ONNX
Object detection
Image segmentation
OCR and document classification
Roboflow and CVAT labeling
Software and cloud
Full-stack, cloud to phone.
Python
TypeScript
JavaScript
SQL
React
Next.js
Node.js
iOS and Android
React Native
Flutter
Google Cloud (Cloud Run, Pub/Sub, BigQuery, Vertex AI)
Firebase
AWS
Azure
Docker
Kubernetes
Terraform
GitHub Actions
OpenTelemetry
PostgreSQL
Firestore
Redis
MongoDB
Design, UX, and research
Find out what people need, then design it.
See Service Design and UX & Research Leadership.
Figma
Prototyping
User interviews
Usability testing
Card sorting and tree testing
Journey mapping
Service blueprinting
Cooper School method
Design systems
Accessibility
Heuristic evaluation
Miro and FigJam
Video, animation, and games
Stories that move, and games people play.
Video production
Animation direction
Adobe After Effects
Adobe Premiere
Final Cut Pro
Motion graphics
Adobe Photoshop
Game design
Godot
Game AI and pathfinding
Level design
Character animation
Gamification
Data and analytics
Know what is working.
Pandas and Jupyter
BigQuery
Tableau
Power BI
Looker
Google Analytics 4
Hotjar
Mixpanel
Amplitude
A/B testing
dbt and Airflow
Geospatial mapping (KML)
Enterprise platforms
The big systems you already pay for.
Adobe Experience Manager (AEM)
Sitecore
Salesforce
ServiceNow
WordPress
Webflow
Retool
Questions
Can you build hardware prototypes, not just software?
Yes. Device prototypes on Arduino and Raspberry Pi, sensor integration, interactive installations, body tracking and gesture recognition, and computer vision running at the edge on NVIDIA Jetson and Raspberry Pi.
Which AI models do you work with?
Claude, OpenAI, and Gemini through their APIs, plus open-weight models such as Llama and Hermes. The model is picked per job.
Is this every technology you have ever used?
No. This page lists what is in use now. Older languages and platforms from the Flash and early web years show up in the portfolio, not here.
Got a problem that needs one of these?
A 30-minute call, no deck, no pitch. Tell me what you are trying to do and we work out whether this toolkit fits.