Skip to content

Careers

We're building the team
that will define AI infrastructure

SRS ResearchHub is an early-stage research organization working on foundational problems in AI infrastructure. We do not have a playbook. We have a rigorous methodology, a growing body of research, and a very hard set of open questions.

What We Look For

We do not screen for pedigree. We screen for how people think.

01

People who think about AI differently — not as a black box to prompt, but as a system to understand at every layer

02

Deep technical curiosity — the ability to hold a problem open for weeks before converging on an answer

03

Systems thinking — understanding how a change in one layer ripples through the entire architecture

04

Bias toward formalization — turning intuitions into theorems, turning experiments into measurable claims

05

Comfort with early-stage ambiguity — there is no playbook for what we are building

06

Intellectual honesty — the ability to say "we do not know yet" and treat that as a research agenda, not a failure

Research Areas

We work across six interconnected research areas. Contributions can span one or several.

01

Information Theory

Entropy bounds, semantic compression limits, lossless vs. lossy tradeoffs at the meaning layer. Shannon gave us the foundation — we are building the next floor.

02

Computational Linguistics

How meaning is distributed across tokens. Which tokens carry semantic weight, which are syntactic scaffolding, and how that distribution shifts across domains.

03

Systems Architecture

End-to-end AI infrastructure design — routing pipelines, priority scheduling, latency-quality tradeoffs, edge deployment constraints.

04

Mathematical Formalization

Turning research intuitions into proofs. Convergence guarantees, optimality bounds, formal verification of system behavior across parameter spaces.

05

Meta-Learning & Self-Improvement

Systems that evolve their own measurement parameters. Autonomous discovery of new dimensions for evaluating information — without human specification.

06

LLM Evaluation & Benchmarking

Designing evaluation frameworks that measure what actually matters — not proxy metrics. Reproducible benchmarks with honest uncertainty quantification.

Technical Skills

Depth matters more than breadth. Strong in two or three areas beats surface-level coverage across all.

Languages

  • Python (primary)
  • TypeScript
  • LaTeX for proofs

Technical

  • Mathematical proof writing
  • LLM evaluation design
  • Statistical analysis
  • Information-theoretic reasoning

Systems

  • API design
  • Distributed systems fundamentals
  • Edge/cloud tradeoffs
  • Performance profiling

Research

  • Experimental design
  • Literature synthesis
  • Hypothesis falsification
  • Academic writing
NOTE

All positions are currently stealth

We are operating in stealth mode. Specific project details, research scope, and technical focus areas are shared under NDA with candidates who pass initial screening. If what you have read here compels you, that is a sufficient signal to reach out.

Ready to work on hard problems?

Send us a note describing how you think about one of the open questions in our research — or a problem in AI infrastructure you believe is underexplored. We read everything.

SRS ResearchHub — Rethinking the fundamentals.