Insights
Published. Shipped.On the record.
The research we publish at ICLR, NeurIPS, USENIX and ACM, and the production systems we ship and keep running.
Zero PII leaks.
Across all tested queries in the ~100-worker MVP.
A PII-safe natural-language data assistant
- LangGraph
- BigQuery
- Vertex AI
Efficiency up 20%.
Cost down 5%, defect rate down 2-3%.
Real-time computer-vision inspection at the edge
- NVIDIA Jetson
- Defect detection
200K+ repriced, every cycle.
Across three vehicle markets, weekly and monthly.
Anchored neural forecasting across three vehicle markets
- Time-Series ML
- Databricks
- MLOps
100% personalized.
Every listing in the portfolio, generated on demand.
Generative personalization for property discovery
- Generative AI
- NLP
Identity held, frame to frame.
Cleaner tracks that hold up under dense, real-world motion.
Advancing multi-object tracking with computer vision
- PyTorch
- IDF1 · MOTA
Conservation at the core.
Recommendations built for water-efficient landscapes and rain gardens.
A warehouse-native recommendation engine
- Snowflake
- Rain gardens
25,000+ changes, each traced to its step.
Fewer than 3% of pairs need a domain expert.
Finding the revisions that actually change the repair
- LLMs
- Databricks
- MLflow
28 stats per team, straight off the TV feed.
Every player in shot placed on a to-scale map of the pitch.
Turning a TV broadcast into tactical match data
- Computer Vision
- Object Tracking
- Sports Analytics
Work with us
Have a problem worth publishing?
If it's hard enough to end up in this record, it's the kind of work we want. Tell us what to build, and we'll show ours.