Radical Reads

Automating traditional data science: Why we invested in Delphina

By Rob Toews, Partner


It is a simple reality: much of what knowledge workers do every day, across industries and across functions, is routine, predictable and formulaic. One of the clearest commercial opportunities for large language models (LLMs) today is to streamline and automate broad swaths of white-collar work.

As an example, GitHub Copilot and similar products have powerfully illustrated the productivity gains that can be unlocked by applying LLMs to software engineering.

Similar to software engineering, “traditional” machine learning and data science represent complex but learnable activities that today’s language models are well-suited to tackle.

Delphina aims to do for traditional ML and data science what GitHub Copilot is doing for software engineering.

In early-stage investing, nothing matters more than the team. When we met Delphina cofounders Jeremy and Duncan, we knew that we wanted to back them and be in business with them. Venture capitalists like to talk about “founder/market fit.” Jeremy and Duncan are founder/market fit, personified. They both have deep, hard-won expertise in data science and machine learning practices and workflows, meaning that they understand as well as anyone on earth how to go about transforming those workflows using generative AI.

Jeremy led the machine learning infrastructure team at Uber for many years, founding and running Uber’s famed Michelangelo platform. He then cofounded Tecton, a unicorn “MLOps” startup commercializing feature stores.

After finishing his PhD at Harvard, Duncan likewise worked as a senior data science leader at Uber for nearly five years before joining GoPuff to lead its data science and data engineering teams.

Jeremy and Duncan’s vision for Delphina is to leverage LLMs to make it dramatically faster and simpler to carry out a wide range of core machine learning use cases—think forecasting, personalization, pricing, or fraud detection, to name a few.

By democratizing the ability to do sophisticated data science, Delphina will drive immense productivity gains and value creation for any organization, large or small, that deals in data—which, in 2023, means pretty much every organization on earth.

Today, we are thrilled to announce that Radical is co-leading Delphina’s $7.5 million seed round alongside our friends at Costanoa Ventures and an esteemed group of angel investors including deep learning pioneer Fei-Fei Li, Weights & Biases CEO/cofounder Lukas Biewald and Nobel Prize-winning economist Guido Imbens.

If you are interested in Delphina’s mission to change the way that data creates value for humanity, see here for the company’s open roles. See here for today’s Bloomberg coverage.

Onward! 🚀

Radical Reads is edited by Leah Morris (Senior Director, Velocity Program, Radical Ventures).