About

Muhammad Farooq

I'm a researcher and builder at heart, working on agentic and generative AI systems. I earned a PhD in applied machine learning and I'm a Google Developer Expert in AI/ML.

Interests. I'm interested in building scalable agentic systems for retrieval and analytics, with a special interest in open-weight models. Earlier, my research was in machine learning applications in wearable systems and biomedical signal processing.

Open source. Most of my work is open source. I created localGPT, one of the earliest fully private, on-device retrieval systems and now one of the most widely used, which lets you chat with your own documents without anything leaving your machine. I also built Verbi, a provider-agnostic speech-to-speech system. I'm currently building whryte.

Frontier models. I work closely with frontier AI labs as an early tester and red-teamer, evaluating unreleased models and providing feedback before they launch.

Industry. As an AI expert, I work with companies to strategize and ship AI solutions, from early design through implementation, with a focus on agentic systems and retrieval. I speak regularly at tech conferences, including Google Cloud Next in 2025 and 2026, where my most recent talk was on building reliable agents. I have also appeared on Google's People of AI podcast. I share what I learn with a large community of engineers on my YouTube channel, Prompt Engineering. Through the channel, I have collaborated on content with companies such as Google, Anthropic, Amazon, and NVIDIA.

Writing and research. My research has been cited more than 2,000 times, with peer-reviewed work published in Scientific Reports, part of the Nature Portfolio. I also write essays on retrieval, reasoning, and local AI here on this site.