Meryem Arik is the co-founder of TitanML, an NLP development platform focused on the deployability of large language models. TitanML helps businesses build smaller and cheaper deployments of language models with ease, tackling the gap between a model that works in a demo and a model that serves real users reliably, and affordably, in production.

Meryem’s path to founding a company was anything but linear. She studied theoretical physics and philosophy at Oxford, describes herself as a recovering physicist, and went from academia into investment banking before realizing that what she truly loves is building products and systems that users genuinely like. She has been in the tech ecosystem ever since. TitanML started as a research project growing out of her cofounder’s research at UCL, and the company was founded before ChatGPT made LLMs a household topic.

That head start shows in the depth of her practical knowledge about deploying LLMs. Meryem speaks frequently on when to prototype with open source models versus API services, and on the optimization work, latency, throughput, and cost trade-offs, that determines whether an LLM deployment succeeds. Her combination of scientific training and product instinct makes her a clear-eyed guide to the realities of putting large language models into production.