Telemetry, machine learning, and cognitive agents.

Tokenization, embeddings, and architecture of LLMs in education.

Zero-shot, few-shot, chain-of-thought, and Socratic prompting.

Connecting vector databases (Pinecone/Weaviate) to courseware.

Capturing student interaction logs across platforms.

Building logistic regression and decision trees for student retention.

Auditing an educational prompt system and predictive model.

Modeling student skill acquisition over time.

Dynamically adjusting question difficulty based on competence.

NLP pipelines for essay scoring and feedback.

Adapting open-source models (Llama/Mistral) on domain datasets.

Securing classroom bots from jailbreaking.

Visualizing model reasoning for educators.

Calculating effect sizes (Cohen's d) post-implementation.

Submitting an architecture for an AI learning platform.