Explore NamasteLM

AI systems, made observable

Understand intelligence.
Build with it.

Interactive AI visualizers, technical learning, developer tools and reference implementations.

Explore NamasteLM View on GitHub
01 / Pipeline

Follow a thought
through the machine.

Move through the stages of language-model generation. Focus or hover on any node to see what happens beneath the abstraction.

02 / The platform

One ecosystem.
Four ways in.

01
Visualize

See how models work.

Explore tokenization, embeddings, attention, transformer layers, sampling and generation through interactive visualizations.

02
Learn

Go beyond tutorials.

Build a rigorous understanding of AI, machine learning and language models from fundamentals through implementation.

03
Build

Learn from working systems.

Study reference architectures, experiments and implementations designed to connect theory with production engineering.

04
Tools

Utilities for AI engineers.

Inspect, analyze, compare and understand models, prompts, tokens and generated content with practical developer tools.

03 / Experiences

Explore the
systems inside.

A growing collection of focused environments for inspecting models, testing ideas and understanding the mechanics of modern AI.

Beta
Tool

Bhasha Lens

Clean up pasted text, copy a reusable version, and inspect Indian-language scripts and token boundaries.

Explore concept
Coming Soon
Visualizer

Transformer Visualizer

Trace representations as they move through a compact transformer.

Explore concept
Experimental
Visualizer

Attention Explorer

Examine the relationships attention heads discover across a sequence.

Explore concept
Coming Soon
Playground

Model Playground

Compare generation settings and observe how decoding changes an answer.

Explore concept
Coming Soon
Tool

Prompt Inspector

Inspect token usage, structure and model-facing prompt composition.

Explore concept
Coming Soon
Tool

Model Utilities

Practical instruments for examining model metadata and behavior.

Explore concept
04 / Learning paths

Knowledge that
compounds.

Structured paths that connect mathematical foundations, model architecture and the realities of deploying intelligent systems.

Path 01

Foundations

  1. 01 AI Fundamentals
  2. 02 Machine Learning
  3. 03 Neural Networks
Path 02

Language Models

  1. 01 Transformers
  2. 02 LLMs
  3. 03 Fine-Tuning
Path 03

Production Systems

  1. 01 Inference
  2. 02 Quantization
  3. 03 Evaluation
  4. 04 Edge AI
05 / Reference projects

Theory meets
a working system.

01

Build a tokenizer

From normalization to vocabulary and byte-pair merges.

  • NLP
  • Algorithms
02

Build a tiny transformer

Implement the essential architecture one block at a time.

  • PyTorch
  • Attention
03

Run an LLM locally

Understand formats, runtimes, memory and hardware tradeoffs.

  • Inference
  • Systems
04

Fine-tune a small model

Prepare data, train adapters and evaluate behavior deliberately.

  • Training
  • Evaluation
05

Quantize a model

Trade numerical precision for practical speed and footprint.

  • Optimization
  • Edge
06

Deploy inference at the edge

Design a lean, observable model-serving path.

  • Cloudflare
  • Deployment
06 / Philosophy

Understanding comes
before abstraction.

Modern AI tooling often hides the systems underneath.

NamasteLM exists to make those systems observable, understandable and buildable — from the first token to the final output.

07 / Open development

Learn openly.
Build openly.

Reference implementations, experiments and selected tools will be developed in public—so ideas can be inspected, challenged and improved.

Explore on GitHub