TUTOR

M30 · Level V — Frontier Intelligence & Research

Intelligent Systems Capstone

Integrate M01-M29 into one independently engineered intelligent system that is specified, implemented, measured, stress-tested, documented and defended.

Topics

  • Problem
  • System
  • Model
  • Data
  • Simulation
  • Decision
  • Verification
  • Deployment
  • Research
  • Defence

Technical concepts

  • 01Baseline Before Breakthrough
  • 02Experiment Matrix
  • 03Evaluation Contract
  • 04Statistical Evidence
  • 05Model Selection and Representation
  • 06Retrieval, Memory and Knowledge
  • 07Agentic and Tool-Using Systems
  • 08Sequential Decision and Control
  • 09Simulation and Digital Twin Discipline
  • 10Scientific and Physics-Aware Intelligence
  • 11Computer Vision / Multimodal Perception
  • 12Language, Speech and Generative Behaviour
  • 13World Models and Internal State
  • 14Continual and Adaptive Intelligence
  • 15Reliability Engineering
  • 16Security and Trust Boundaries
  • 17Performance and Compute Engineering
  • 18Human Oversight and Interaction
  • 19Architecture Review Gates
  • 20Reproducibility Contract
  • 21Repository Architecture
  • 22Logging and Provenance
  • 23Research Literature Matrix
  • 24Capstone Lab 01 - Problem Framing
  • 25Capstone Lab 02 - Baseline Construction
  • 26Capstone Lab 03 - End-to-End Intelligence Loop
  • 27Capstone Lab 04 - Ablation
  • 28Capstone Lab 05 - Robustness Campaign
  • 29Capstone Lab 06 - Fault Injection and Recovery
  • 30Capstone Lab 07 - Resource and Performance Profile
  • 31Capstone Lab 08 - Scientific Validation
  • 32Capstone Lab 09 - Security / Adversarial Evaluation
  • 33Capstone Lab 10 - Independent Reproduction
  • 34Capstone Milestone Plan
  • 35Architecture Decision Record (ADR) Rules
  • 36Verification and Validation Matrix
  • 37Safety Case Structure
  • 38The Failure Taxonomy
  • 39Claim-Evidence Chain
  • 40Scientific Communication
  • 41Technical Defence Format
  • 42Final Deliverables
  • 43Release Readiness Checklist

Tools and technologies

  • Linux VM/container
  • versioned repository
  • reproducible runtime
  • tracked datasets
  • experiment registry simulation
  • evaluation harness
  • technical defence

Expected learning outcomes

  • 01Every major claim must connect to an explicit requirement, mechanism, measurement, evidence artifact and limitation.

Assessment

  • 01Problem and specification — 10%
  • 02Technical integration — 20%
  • 03Experimental design — 15%
  • 04Empirical evidence — 15%
  • 05Reliability/security — 10%
  • 06Research quality — 10%
  • 07Reproducibility — 10%
  • 08Defence — 10%

Source document: TUTOR Frontier Module 30 Intelligent Systems Capstone