Sorting 75 Years of AI/ML and Software Concepts for Project and Venture Success Hi friends, a major problem today is that founders jumping into AI require a crash course on AI/ML fundamentals to organize knowledge and strategies like experts. For example, stakeholders may confuse concepts like RAG, where the model is fetching embedded data and feeding it in with a prompt, with fine-tuning where custom org data is used to update the weights of an LLM. Our aim is to democratize access to the fundamentals with the following resources. • AI/ML Primer for STEM People: Analogically map popular AI/ML concepts, learning them at a high level. • Topological Sort of AI/ML Models, Algos, and Strategies: Zoom through 75 years of AI/ML history from the basics of CS to the most advanced transformers. Cover a 13-step process from data prep to training to inference to online learning. Explore trending APIs, infrastructures, and data sets. • Learn about important terms like embeddings, regression, BN, SVM, Dim. Red., NN, CNN, RNN, BERT, GPT, GAN, fine-tuning, RAG, stable diffusion, PI/VI/DQN, Q Learning, SGD, PGM, Adam, ReLU, Loss Functions, Actor-Critic, Few Shot Learning, Prompt Engineering, Multi-head Attention, etc. all in one sheet. • Meta dogfooding: To create the canvas, we used AI/ML techniques, including GPT, to extract, organize, and generate content from tens of thousands of pages of AI/ML course materials and API documentation. The information is presented in a topological order (using GPT-generated DAGs of both topics and subtopics), ensuring a structured learning path. where you don’t feel you lack prerequisites. The process involved embedding concepts, clustering them by cosine similarity, and using prompt-based and actor-critic strategies for summarization and refinement. • AI/ML Terms for Advanced Practitioners: Here are all the terms across AI, ML, DL, PGM, NLP, CV, RL, Robotics, etc. that we believe experts use to talk to each other. Let us know if we are missing any. • Project Success Canvas: The tabs above are part of a greater Project Success Canvas. Over the years, we have worked with 100+ companies and seen many succeed and many fail. We document the software best practices across web, mobile, UIUX, digital marketing, SEO, DevOps/CI, software design patterns, etc. Like the AI/ML tab, while all the information may be online, this is a single go-to place for everything. We also include tidbits about organizing codebases and legal ramifications of global collaboration. • Ventures Canvas: While many business model canvases exist, few fully incorporate design research, entrepreneur mental health, idea banks, etc. into the process. Of course, we build off the giants, referencing the likes of YC Startup School, MIT EdX Entrepreneurship, Techstars Toolkit, UIUC iMBA, Crunchbase startup finance, etc. My coauthors: Sugam, Nsamba, Sampanna, and others listed on the canvas. Here's my LinkedIn as well. We're proud to say this canvas was made mostly by AI steered by technologists in Africa and Nepal.