DataFrontier

Careers

Principal Data Scientist — ASOC and Persistent Memory Substrate

Set technical direction across ASOC and the persistent memory substrate that future UPII will be built on.

About the role

DataFrontier has two hard technical problems that are converging. ASOC is an AI revenue operating system whose agents reason over heterogeneous, real-time signals and hand off to humans at the decisive moment. UPII is the research program for a personal-intelligence substrate — memory, identity, and permissioned reasoning that serve the individual. The persistent memory substrate sits underneath both.

As Principal Data Scientist, you will set technical direction across that surface. You will decide how agents reason over signal, how memory is structured so it survives beyond a context window, and how the substrate stays correct under real production load. This is the most senior individual-contributor research role in the lab.

You will not run a team in the management sense; this role has no direct reports. You will own problems — the ones whose answers change what we build. You will work directly with the founders, prototype the approaches yourself, and set the bar for how the rest of the team works.

We are looking for someone who has done this before — who has taken an ambiguous, high-stakes problem and turned it into a system that works, and who can write down why it works so others can extend it. If you want to set direction rather than execute someone else's, this is that role.

What you'll do

  • Set technical direction — across ASOC's agent reasoning and the persistent memory substrate.
  • Solve the memory problem — structure memory so it persists and stays correct beyond the context window.
  • Prototype yourself — you build the approach, not just specify it.
  • Define evaluation — decide how the lab knows whether a system actually works.
  • Raise the bar — set the standard the rest of the team builds to.
  • Write it down — publish the results that should outlive the project.
  • Work with the founders — on the questions that change the roadmap.

Who we're looking for

  • Principal-level track record — you have set direction on hard problems before.
  • Depth in ML and systems — you are credible on both the model and the machine.
  • Memory and retrieval — you understand why long context is not the same as memory.
  • Agent systems — you have shipped LLM agents that run in production.
  • Evaluation rigor — you measure honestly, including the failures.
  • Strong writing — you can make a complex result legible to others.
  • Judgment under ambiguity — you choose what not to do as well as what to do.
  • Still hands-on — you have not left the keyboard.
  • Self-direction — you set your own week and own the outcome.
  • Relevant background — in one of our core domains — revenue/agent systems, personal-data substrates, or large-scale retrieval.

Nice to have

  • Published research — papers or technical notes the field cites.
  • Personal-data substrates — direct experience with the UPII problem space.
  • Production agent orchestration — you have run multi-agent systems at scale.
  • Cryptography or provenance — keys, capabilities, attribution.
  • Mentorship — you have raised the level of people around you.
  • Open-source leadership — you have stewarded something others depend on.

How we hire

  • A short application: CV, links, two writing samples, and a note on a problem in this space you would want to own
  • A 60-minute conversation with both founders on direction and fit
  • A paid take-home problem, scoped to ~8 hours, on a real question
  • A half-day on-site in Bangalore, including a technical deep-dive
  • Reference conversations with people you have worked with
  • An offer within 7 working days of the on-site, or a clear no with feedback

We do not ghost. Whether or not we move forward, you will hear from us at every stage.

What we offer

  • Hybrid: 3 days in office (Tuesday/Wednesday/Thursday at HSR Layout, Bangalore), 2 days remote
  • Meaningful equity, calibrated to seniority and discussed at the offer stage
  • A voice in who we hire and how the team works
  • Annual research budget for books, conferences, courses, and compute — your call on how to spend it
  • Conference attendance and publication support at the level this role works
  • Health insurance for self, spouse, two children, two parents
  • 28 days of paid time off, plus public holidays
  • Compensation is calibrated to seniority and prior outcomes, discussed at the offer stage