Principal Research Scientist (Computer Science / Computational Physics)

What you'll do

Where the junior version of this role runs the experiments, this one decides which experiments are worth running and what the results mean. You frame foundational open problems, theoretical bounds or simulation challenges where progress can actually be verified, then guide AI models and agents through multi-step research (proofs, algorithms, numerical experiments, high-performance simulations). You validate the output with proof checks, computational reproduction and formal methods, specifically to find where model reasoning breaks down, and you rate novelty, correctness and significance against current literature.

The written output is research memos on attempts, partial solutions and model failure patterns, plus advice on which subfields and open problems are most likely to yield to AI-assisted work. micro1 frames the role as defining what counts as genuine progress in your area.

Remote contract work, part-time or full-time.

Who fits

micro1 describes a senior, field-level profile (all listed as preferred):

• 7–10+ years of active research in computer science or mathematical or computational physics
• Field-level recognition: major prizes, society fellowships, plenary or invited talks at top venues
• A steady record in leading venues, with results published within the past year
• A track record of resolving named open problems or introducing concepts others have adopted
• Deep hands-on computational experience: simulation, numerical analysis, algorithm implementation or HPC
• The ability to critique rigorous work outside your own specialism
• Strong plus: program committee or editorial board service, formal tools such as Lean, Coq or SAT/SMT solvers, or a public record of posing open problems or doing research with advanced AI systems

No prior AI experience is required. The work is in English.

What it pays

micro1 lists $100–400/hour, the same band as the earlier-career Research Scientist posting despite the much higher bar. The ad says nothing about how placement in the band is decided or whether pay is per hour or per output, so a principal-level researcher should ask where in the range an offer sits before committing.

Worth knowing

Good:

• An advisory, direction-setting role rather than production grading
• High ceiling and a flexible commitment, part-time or full-time
• Fully remote, with no country restriction stated
• Studying where frontier models fail at research is interesting work in its own right

Less good:

• Four openings for a profile few people meet
• Same published band as the junior listing, so seniority is not visibly priced in
• No compensation structure, weekly hours, screening steps or start date published
• The customer, and whether research credit or publication is possible, are not stated

About this listing

We read this ad on micro1's signed-in jobs board on 18 October 2026, six days after it was posted. Pay, openings and qualifications are micro1's own text; the gaps noted above are gaps in the ad itself. See micro1.