Learning
Concepts tied to evidence. No arbitrary skill percentages.
With evidence
Agents
An evaluable action loop, not a chatbot with plugins.
Data engineering
Pipelines, lake, warehouse. 2021 base: Sparkify training, not a client.
Evaluation
Metrics first, interpretation second. No arbitrary LLM scores.
Experiment tracking
Versions, params, artefacts. Without them you cannot compare two runs.
RAG
Retrieval before generation, visible citations, explicit uncertainty.
Tool use
The tool runs, the model reads the result. No silent field invention.
Exploring
MCP
A tool protocol for agents. Exploring: no public MCP server in this Lab.
ROS 2
Robot middleware. Exploring: no public ROS 2 workspace in this Lab.
Simulation
Synthetic world before the real one. Exploring: no published robot or physics sim.
VLA
Vision-language-action. Exploring: reading, no deployed robot policy.
VLM
Vision-language models. Exploring: vocabulary and reading, no public fine-tune.