We research data engineering, AI/ML systems, and how model capabilities transfer across languages and domains — then ship what we learn into MPP BI and Lizardata.
Latest from the lab
What we measured, what broke, and what we’d do differently — written up as we go.
Testing Nemotron 3 Super in MPP BI: a single-GPU 120B model in practice
We loaded NVIDIA's 120B Nemotron 3 Super on a single GPU and tested it inside MPP BI against GPT-OSS-120B, on speed, long context, and Armenian and Georgian reasoning.
Does the language of a spatial instruction interface affect an LLM agent's ability to act?
We tested whether the language of an LLM agent's operating interface, not just its goal, changes how well it completes a multi-step spatial task. Translating the ongoing interface hurt performance far more than translating the one-time goal.
Does context make AI analytics more efficient?
We asked an AI agent to build dashboards on the same data, with and without a data catalog and glossary. Median token use fell 14.2%, and passes rose from 77.8% to 96.3%.
Built by the team that does the research.
The people who publish the Lab Notes are the ones who build your pipelines, tune your models and run our products in production.
Research you can run.
Industries
The same research and the same products, applied to the data each industry already runs on.
Talk to us, or see the research first.
Bring the data problem you have, or start with what we’ve already published. Either way, you’ll be talking to the people who did the work.