Computer Architecture Researcher (Research Coding Tasks)

$80/hr

rate as listed

Workload: ~15h/week
Engagement: Project-based
Location: Worldwide
Field: Engineering

Convert published computer architecture research and released code into coding tasks for frontier AI agents. The work includes designing hidden evaluations, setting metric thresholds, preparing reference solutions and reviewing agent attempts.

The work

You will start from a recent computer architecture paper with released code, ideally your own. You will remove the implementation of its central contribution, preserve the surrounding simulator and benchmark infrastructure, and write a task that asks an AI agent to reconstruct the missing work and improve on the paper's reported results.

Evaluation design

Each task requires held-out workloads, per-metric pass thresholds, a hidden reference solution and a second solution demonstrating that the stated bar can be exceeded. Later, you will review agent submissions for correctness, reproducibility and whether apparent gains reflect the intended architectural improvement rather than benchmark leakage or simulator artifacts.

Screening

Afterquery screens for published computer architecture research with released code and a completed or in-progress PhD in computer science or a related field. Relevant depth includes caches, prefetchers, branch prediction and accelerators evaluated with simulators such as gem5 or ChampSim. First-author work with measured improvements, Docker experience and reproducible benchmark setups are preferred, and the application process includes background checks.

Logistics

The contract is fully remote, asynchronous and flexible. Each task typically takes a few hours, although the exact effort depends on the paper, simulator and evaluation setup.

Pay band: $80/hr

What they look for

Published computer architecture research with released implementation code
PhD in computer science or a related field, completed or in progress
Computer architecture experimentation using simulators such as gem5 or ChampSim
Held-out benchmark design and per-metric threshold setting
Reproducible research coding and benchmark packaging
Technical review of AI-generated implementations

Included with this listing

Interview prep — 10 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: This tests whether you can ground your candidacy in a verifiable publication and explain its experimental claim precisely enough to support a task specification.
A strong answer: Identify the paper and venue, state the architectural mechanism, name the baseline and workloads, and report the relevant metrics and comparison conditions. Clarify your authorship role and which parts of the implementation you personally handled.
Where people lose it: Candidates lose points by describing a broad research area without naming a paper, baseline, benchmark suite or measured improvement.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.