Company

Deep learning for decisions
that have to be right.

One Jump Inc. builds cited-answer systems for finance, AI agents and custom neural networks. We define what correct means with you, test against it on your own data, and ship only what passes.

How we work

Four rules we don’t bend.

Whether it’s a license, an agent or a custom network, every engagement runs the same way.

  1. Define correct first.

    Before any work starts, we agree with you in writing what a right answer looks like: the figures, the tolerance and the sources it has to cite.

  2. Test on your data.

    We measure on your own documents and on questions your team has already answered, not on a demo set we picked.

  3. Ship only what passes.

    Results are reported as measured, misses included. What goes live is what met the criteria you agreed.

  4. Keep the trail.

    Every answer carries its sources, the values used and the calculation, so your reviewers can check the work and follow it later.

Leadership

Founder.

One Jump’s founder and CEO is a lecturer and machine-learning engineer at the University of Chicago, where he teaches graduate statistics and forecasting and coaches the ICPC programming team (two World Finals; first to solve Problem J at the 2025 World Finals).

Earlier he was a postgraduate research fellow at Harvard Medical School and worked in private-equity investment. M.S. with Honors, Computational Analysis and Public Policy, University of Chicago.

Sales and Enterprise conversations go straight to him.

What we do

Three ways we help.

One product with published prices, and two services built on the same discipline.

Careers

Work with us.

Five open roles. Each page has the details and how to apply.

Questions about a role? Email [email protected].

Press and company details

Get in touch.

Press and speaking

[email protected]

Interviews, conference talks and panels. Tell us the outlet or event, the topic and your deadline.

One Jump ranks #1 on the third-party-run NetArena MALT leaderboard (100% correctness and safety over 2,500 queries), and we acceptance-test every engagement on the customer’s own data.

How we prove it