Rajistics

Rajistics' real name is Rajiv Shah. He links to his personal site at rajivshah.com across all his social profiles, making the connection straightforward to find. The handle 'Rajistics' blends his first name with a data-analytics twist that suits his machine learning and data science content.

Rajistics used The Simpsons as a fun, recognizable dataset to teach how Mixture of Experts models work from scratch, reflecting his stated goal of making ML content that is both informative and occasionally funny. It's a deliberate style choice — using pop-culture material to make a complex modern architecture feel less intimidating. Several large language models in use today rely on the Mixture of Experts approach, so the topic is genuinely relevant beyond the humor.

Yes — one of his recent videos walks through evaluating agentic AI skills using OpenHands, an open-source framework built for AI agent workflows. The focus is practical evaluation methodology, which fits his broader theme of helping developers and researchers actually measure what their AI systems can do rather than just build them.

Yes, Rajistics has a dedicated video on Graph RAG for code using a tool called GitNexus. Graph RAG combines knowledge graphs with retrieval-augmented generation to improve how AI systems reason over structured information, and applying it to codebases is a growing area of interest in the developer AI space.

Hugging Face is one of the most consistent topics across Rajistics' content, appearing regularly in his hashtags and video subjects. His tutorials on training models, evaluating large language models, and working with open-source AI tools naturally lean on the Hugging Face ecosystem, making the channel a useful resource for practitioners already familiar with that platform.

Evaluating generative AI is a central theme on the channel — Rajistics has produced and actively updated a practical guide to evaluating GenAI applications, with a major revision as recently as late 2025. He covers concepts like functional correctness, unit-test-style evaluation for code models, and retrieval-augmented generation quality, going well beyond basic benchmarks to address real-world deployment concerns.

Most of Rajistics' material targets intermediate-to-advanced practitioners — topics like training Mixture of Experts models from scratch, Graph RAG for code, and agentic AI evaluation all assume baseline familiarity with machine learning. His practical-guide format and occasional humor do lower the barrier compared to academic papers, but someone completely new to ML would likely find the content challenging without prior foundations.

Yes, Rajistics lists Medium among his active platforms alongside YouTube, TikTok, Instagram, and Twitter, all under the @rajistics handle. He appears to use written articles on Medium to complement the video tutorials on his channel, which is common among technical educators who want to reach both readers and viewers.

Beyond YouTube, Rajistics is active on TikTok, Instagram, Twitter, and Medium, all under the @rajistics handle. He also maintains a personal website at rajivshah.com that serves as a central hub for his work across machine learning, data science, and AI.

Rajistics explicitly includes the broader impact of AI in society as part of his stated channel scope, alongside technical topics like ML techniques and the latest AI news. This means the channel occasionally steps back from pure tutorials to discuss issues facing the AI community more widely, which broadens its appeal beyond just engineers writing code.