Omkar Ray

Long-form explorations into topics I find fascinating.

Structured thinking on themes I'm obsessed with.

Shorter thoughts, observations, and notes from the field.

Themes I track closely at the intersection of finance, regulation, and AI-native products.

Core machine learning areas I actively study, build with, and write about.

• Transformers Attention Is All You Need, and Why
• Reinforcement Learning Learning from Reward, Not Labels
• AI Agents Agents That Plan, Act, and Recover
• LLM Evaluation When Benchmarks Lie: The Maths of LLM Evals
• RAG Retrieval-Augmented Generation Done Right
• Multimodal AI One Model, Every Modality
• Gradient Descent The Geometry of Gradient Descent
• PyTorch Tensors to Multi-GPU: Visual Map

Mathematical foundations that shape how I reason about models and systems.

• Bayesian Optimization Probabilistic search & surrogate models
• Gaussian Processes Distributions over functions
• Tensors Multilinear algebra & geometry
• Hilbert Spaces Infinite-dimensional inner product spaces
• Manifolds Geometry beyond flat space
• Probability Theory Measure-theoretic foundations
• Linear Algebra Vectors, matrices & transformations

Visual artifacts — packaging concepts, interface explorations, and design systems.