Neoklis Labs
An artificial intelligence research and product company.
Bridging the gap.
We're building a future where everyone has access to the knowledge and tools to make AI work for their unique needs and goals.
While AI capabilities have advanced dramatically, key gaps remain. The scientific community's understanding of frontier AI systems lags behind rapidly advancing capabilities.
Knowledge shouldn't be concentrated.
Knowledge of how these systems are trained is concentrated within the top research labs, limiting both the public discourse on AI and people's abilities to use AI effectively.
And, despite their potential, these systems remain difficult for people to customize to their specific needs and values.
To bridge the gaps, we're building Neoklis Labs to make AI systems more widely understood, customizable and generally capable.
Research & Writing
Insights and deep dives from the Neoklis Labs team.
Deriving Muon
An in-depth look at Muon, an optimizer specifically designed for Linear neural network layers. Learn how normalizing weight updates can modularize deep learning theory.
A Visual Guide to Mixture of Experts (MoE)
Explore Mixture of Experts (MoE) through 50+ visualizations — covering expert routing, token dispatch, capacity factors, and how MoE powers frontier LLMs.
A Visual Guide to Quantization
A visual exploration of LLM quantization — from float basics and symmetric/asymmetric quantization to GPTQ, GGUF, and BitNet weight-only methods.
A Visual Guide to Mamba and State Space Models
A visual guide to Mamba and State Space Models — from SSM fundamentals and discretization to selective state spaces and how Mamba challenges the Transformer.
Validating your Machine Learning Model
A comprehensive guide to ML model validation — from train/test splits and k-fold CV to nested CV, LOOCV, and statistical model selection tests.
Unit Testing for Data Scientists
A practical guide to unit testing for data scientists — covering pytest, fixtures, mocking, and test-driven development for ML pipelines.
Dimensionality Reduction with Principal Component Analysis
A deep dive into the mathematical foundations of Principal Component Analysis (PCA) for dimensionality reduction.
On-Policy Distillation
How on-policy distillation combines the reliability of on-policy training with the dense reward signal of distillation — achieving frontier performance at a fraction of RL cost.