Skills
From the pedagogical model to the inference stack, from the first spec to the stage.
Pedagogical AI and instructional design
ITS architecture
Domain, learner and tutoring models, and the interface between tutor and learner.
Adaptive learning
BKT, FSRS, IRT, PFA, KST and misconception diagnosis.
Learning design
Objectives, sequencing, positioning tests and remediation paths.
Agents and harness engineering
Agent theory
Wooldridge's agency: autonomy, reactivity, pro-activeness and social ability.
Harness design
Tool governance, permissions, sandboxing, traces and guardrails.
MCP and orchestration
MCP server design, tool taxonomy, evals and multi-agent orchestration.
Inference and model selection
Model selection
Choosing the right model per task across quality, cost, latency and privacy.
Serving
Self-hosted inference with vLLM, and open-weights versus API trade-offs.
LLMOps
Gateway and routing with LiteLLM, tracing and evals with Langfuse.
Product and delivery
Product discovery
Specs, PRDs, scoping and acceptance criteria.
Project governance
Steering committees, project reviews, demos and workshops.
Platform
Docker, CI/CD, observability, incident response and post-mortems.
Speaking and training
Talks
Keynotes on AI and learning, LLM-based Intelligent Tutoring Systems and pedagogical harnesses.
Workshops
For training organisations, universities and EdTech teams.
AI literacy
Upskilling for trainers, instructional designers and L&D leaders.