Find a concept

Search Notes and Discovery

Enter a term to search published Notes and Discovery.

    Sai AI Knowledge Hub

    Build durable intuition for modern AI.

    Visual explanations for the concepts that last, plus concise field notes on what is changing now.

    Topics
    5
    Durable Notes
    10
    Discoveries
    3

    Durable understanding

    Explore Notes

    Follow intentionally ordered paths instead of piecing together isolated posts.

    foundation

    Transformers

    A structured path through attention, multi-head composition, positional information, and encoder-decoder architecture.

    Notes
    4
    Range
    Foundation–Intermediate
    foundation

    Tokenization

    From text and bytes to subword algorithms, domain tokenizer training, held-out evaluation, and model compatibility.

    Notes
    1
    Range
    Foundation
    foundation

    Prompt Engineering

    Design, test, and improve prompts through worked examples, structured outputs, reasoning strategies, and inspectable workflows.

    Notes
    3
    Range
    Foundation–Advanced
    foundation

    Classical System Design

    Understand how networked systems communicate, scale, and stay reliable through concrete requests, worked examples, and design trade-offs.

    Notes
    1
    Range
    Foundation
    foundation

    Fine Tuning

    Adapt pretrained Transformers to new tasks through clear terminology, worked metrics, and complete encoder and decoder training workflows.

    Notes
    1
    Range
    Foundation

    A clear first path

    Start Here

    Begin with the map, then add the mechanism. Each step is ordered so the next idea has somewhere to land.

    1. Transformers · foundation

      Attention mechanism in detail

      A step-by-step account of queries, keys, values, masking, scaling, softmax, and the final weighted sum.

    2. Transformers · intermediate

      Multi-head attention

      How parallel attention projections create complementary views and recombine them into one representation.

    3. Transformers · intermediate

      Positional encoding and the encoder-decoder

      Where position, masks, residual paths, encoder blocks, and decoder blocks fit in a complete Transformer.

    4. Transformers · intermediate

      Transformer Encoder — From Scratch with PyTorch

      Follow five tokens through a manually implemented PyTorch encoder, with equations, runnable code, tensor outputs, and a complete dimension audit.

    From the field

    Recent Discovery

    A small, chronological feed of ideas worth carrying into deeper study.

    Transformers ·

    Autoregressive models, step by step

    How next-token factorization, causal masking, training, and decoding turn a prefix into a generated sequence.

    2 min read

    Behind the notes

    A notebook that shows the work.

    This knowledge hub turns ongoing AI study into explanations that can be checked, connected, and revisited—without pretending every open question is settled.

    Built by Sai Prasanna Maharana, a software engineer focused on useful, explainable AI systems.

    About Sai