APPENDIX H: RESUME PORTFOLIO BUILDING CAPSTONE PROJECTS

AFTERWORD You have reached the end of the book, and you started with no background at all. Along the way you learned that a model is only the engine, and that the harness around it, with its loop, tools, memory, safeguards, and tests, is what makes the whole machine dependable. You learned seven layers to organize your thinking: execution, tooling, context, lifecycle, observability, verification, and governance. You also built a small agent with your own hands, broke it on purpose, and fixed it, which is the best way to learn the craft. Three habits will serve you more than any single technique. First, keep things simple, and add complexity only when a measurement shows you need it. Second, assume that anything coming from outside is untrusted, and layer your safeguards so no single failure becomes a disaster. Third, measure honestly, and write down what you find, including the gaps. This field moves quickly, so expect that specific tools, versions, and statistics will change while the underlying ideas stay useful for much longer. Keep your lab, keep testing it, and keep reading from the real sources listed in Appendix F. Be honest with the people who use your systems about what they can and cannot do. Take care of the people your agents touch, because reliable engineering is ultimately about trust. Thank you for reading, and welcome to the profession of harness engineering.