AI Research
Intelligence Builders is developing AI-powered tools for education, including automated student performance analysis using both cloud-hosted and locally-run large language models. Our work focuses on generating accurate, evidence-based summaries from gameplay telemetry data collected by the Strata System.
Publications and Reports
LLM Infrastructure Evaluation
Analysis of cloud GPU providers, on-site hardware options, and a recommended hybrid architecture for running open-weight LLMs to generate student performance summaries.
AI Summary Teacher Guide
Guide for teachers on understanding and using AI-generated student summaries in Mission HydroSci, including how summaries are produced and how to interpret them.
Local LLM Test: Qwen3 8B
Results from testing the Qwen3:8B open-weight model running locally via Ollama on Apple Silicon, including prompt design, output consistency, and accuracy analysis.