Rohit Patel·Jul 2The Picks and Shovels of Orbital Compute EconomyBeyond launch services and satellite operators, the real value of the space-based AI boom lies in the specialized hardware and materials
Rohit Patel·Jun 19Technologies needed to make orbital compute realOvercoming the thermal, radiation, power, and communication hurdles required to run the world’s heaviest workloads in Low Earth Orbit.A response icon2A response icon2
InData Science CollectivebyRohit Patel·Aug 16, 2025What does the future of AI look like if we hit the LLM scaling wall?Why small models, scaled inference, and AI agents maybe be itA response icon1A response icon1
InData Science CollectivebyRohit Patel·Aug 10, 2025Understanding reinforcement learning for model training from scratchAn intuitive treatment of RLHF, TRPO, PPO, GRPO, DPO and RLAIF. This article follows my paper here: https://arxiv.org/abs/2509.04501A response icon4A response icon4
InData Science CollectivebyRohit Patel·Jun 15, 2025An intuitive treatment of Negative log-likelihood, Cross entropy, KL divergence, and Importance…For the past few months, I have been working on a follow-up to my earlier article “Understanding LLMs from Scratch Using Middle School…A response icon2A response icon2
InTDS ArchivebyRohit Patel·Oct 19, 2024Understanding LLMs from Scratch Using Middle School MathIn this article, we talk about how LLMs work, from scratch — assuming only that you know how to add and multiply two numbers. The article…A response icon105A response icon105
InQuickAI.appbyRohit Patel·Aug 6, 2024How to do the Price-Volume-Mix waterfall rightBreaking down revenue changes accuratelyA response icon2A response icon2
