← All roles

Research Engineer, Frontier Data

Training Superintelligence· Frontier Data Operations
Apply on company site
Remotefull-timePosted 3mo ago

About the role

Remote research engineering role focused on delivering frontier-quality datasets, RL environments, and evaluations for a frontier AI lab client, with emphasis on data quality, verification, and measurable model lift.

Responsibilities

  • Respond to inbound client requests and translate evolving research goals into technical proposals and data requirements.
  • Provide feasibility assessments, assumptions, acceptance criteria, and evaluation requirements for scoping contracts/SOWs.
  • Define what “good” looks like via rubrics, boundary cases, and counterexamples; conduct deep audits of produced data.
  • Drive iterative improvements by analyzing failures and turning insights into production instructions (better few-shot examples, counterexamples, clearer guidance).
  • Design and build datasets and RL environments across capability areas (coding/SWE agents, RL environments, multimodal data, STEM reasoning, embodied AI/VLM-driven agents).
  • Create and maintain ground-truth signals, verifiers, unit tests, and environment interfaces; support synthetic data generation and augmentation where appropriate.
  • Develop robust validation, denoising, and data-quality systems; implement automated validation, deduplication, leakage checks, and diversity controls.
  • Build documentation and data cards detailing dataset intent, limitations, usage, and evaluation linkage.
  • Design and run evaluations to prove impact; analyze pre/post comparisons and ablations; support in-house fine-tuning or RL experiments when needed.
  • Collaborate with large production teams, providing clear specs and fast feedback; communicate technical findings to clients and align scope/schedule accordingly.

Qualifications

  • 4–5 years of experience building or improving deep learning systems where data quality materially impacted outcomes (training, post-training, evals, or agentic systems).
  • Strong intuition for data ingredients that drive model improvements: what to collect, filter, synthesize, and how to measure.
  • Ability to communicate clearly with researchers and engineers; translate research objectives into concrete specs and outputs into actionable insights.
  • Comfortable being the direct technical point of contact for a client, including explaining setbacks or trade-offs.
  • Demonstrated ability to be extremely detail-oriented in diagnosing subtle data quality issues and failure modes.
  • Solid programming ability with a bias for shipping: Python proficiency required; SQL/structured data workflows strongly preferred; familiarity with a major language (C++, Java, Go, Rust, JS/TS) a plus for coding-focused work.
  • Comfort designing quality systems: rubrics, validation scripts, gold sets, sampling strategies, statistical checks, slice-based evaluation, and human-in-the-loop review loops.
  • Direct experience with a frontier AI lab is not required; hands-on engineering depth and ability to operate with ambiguity are valued.

Skills

Required:PythonCommunication
Nice to have:SQL

Benefits

  • High autonomy and fast iteration; Strategic Project Lead handles scaling and resourcing.
  • Opportunity to work directly with frontier AI labs and contribute to post-training and RL environment design.
  • Potential for showcasing work at leading AI conferences (e.g., ICLR, ICML, NeurIPS).
Last updated Sep 30, 2026. Interested? Apply on the company site.

About Training Superintelligence

Turing develops large-scale RL environments and data generation systems that train multimodal agents to improve model performance in coding, real-world, economically valuable tasks, and advanced STEM reasoning.

Train AIBuild AIHire AI TalentAI for BFSIEMNAIDatasetsEnterprise Knowledge WorkMultimodal
Company profileWebsite

More open roles at Training Superintelligence