Lessons learned from Master's thesis

• Firmly define the research topic.
• Do literature research based on keywords, never limit the scope of journals and conferences, and continue literature research even after starting the project.
• Collect baselines and benchmarks as soon as possible; make sure the baselines and benchmarks are downloaded and can be run.
• Focus on the NOVELTY of the core aspects; don't be perfect, and don't support all cases or possibilities.
• Find a few small examples that prove the power of your novelty, and emphasize it in both the paper and the presentation.
• Avoid designing overly complex rule-based systems.
• With GenAI, aim to get a prototype and an illustrative presentation out as soon as possible. This will allow you to collect feedback early, empowering you to make necessary adjustments and improvements.
• Use functional programming architecture that is easy to test. Conduct thorough unit testing, document the coverage, and leave traces of development step by step. This will provide a solid foundation and instill confidence in the robustness of your research.
• Insert breakpoints where you're unsure how to code. Figure out what to do when encountering them, and refine the unit tests in the process.
• Don't be perfect; start writing your thesis when you're almost done.