Data For Science

About

Data For Science helps teams take LLM, RAG, and agentic systems from a promising prototype to something they can run, measure, and maintain. Senior hands on the work, clear communication with the people paying for it, and a handoff that leaves the capability with you.

Bruno Gonçalves

Bruno Gonçalves

Founder

Bruno Gonçalves founded Data For Science after serving as VP of Data Science & Finance at JPMorgan Chase, where he led ML and LLM deployments for the bank’s finance divisions — systems that had to be right, auditable, and fast. He has taught 170+ live sessions for O’Reilly Media, Pearson, and Packt since 2019, and writes the Data Science Briefing, a weekly newsletter with 340+ issues and 3,000+ readers. Before industry, he was a Data Science Fellow at NYU’s Center for Data Science and earned a PhD in the Physics of Complex Systems (Emory University, 2008); his research has been covered by MIT Technology Review, The Economist, Wired, and The Atlantic.

The through-line is the same in every role: make a complex system measurable, then make it work.

Recent writing

We publish the code behind the advice. The three most recent posts:

  1. Jev 101

    A new class of model

  2. Local LLM Inference at Scale with vLLM

    Continuous batching, prefix caching, and structured outputs: the vLLM features that turn a single local model into a serving engine for thousands of requests.

  3. The Epidemiology of a Cognitive Virus

    What happens when you take "LLM addiction" seriously enough to write down the equations and run simulations

Evals-first engineering

If it can’t be measured, it isn’t done. Every engagement leaves behind a test set, a harness, and numbers you can defend.

Built for handoff

Decision-ready trade-⁠offs for stakeholders, clean code and docs for engineers — so the capability stays with your team, not with us.

Contact

The fastest way to find out whether we can help is a 30-⁠minute call. You leave with written next steps either way.

Email & phone

Prefer email? Send a brief note with your use-⁠case, timeline, and constraints.

Helpful context

  • Your use-⁠case (one paragraph)
  • What “success” means (quality, speed, cost, reliability)
  • Data availability
  • Privacy constraints
  • Timeline and stakeholders

Book a call

30 minutes. We’ll pin down the problem, the success metric, and the shortest path to a working system.

Open scheduler