Research Engineer, Frontier Data
Salary not listed • Remote • Brazil • Full-time
Posted 7mo ago
This job has been open for more than 60 days. It may no longer be actively hiring, so check with the company before putting a lot into an application.
Remote Work Policy
Fully remote
Employment Type
Full-time
Categories
AI Research Engineer
About the job
Turing is seeking a Research Engineer to contribute to the development of frontier-quality datasets, reinforcement learning environments, and evaluations for state-of-the-art models. This role involves close collaboration with client researchers and engineers, translating research goals into technical proposals and data specifications, and overseeing technical execution. You will be responsible for building quality and verification systems to ensure deliverables meet high standards for correctness, realism, diversity, and measurable model lift. The work is bespoke, focusing on a single frontier lab client to build deep context and relationships, with flexibility to support other accounts. Success is measured by both turnaround time and quality. This position is ideal for individuals experienced in building and improving deep learning systems, with a strong emphasis on data quality, curation, denoising, synthetic data generation, and rigorous evaluation.
Responsibilities
- Own data and environment quality from an AI researcher perspective.
- Respond to client requests and translate research goals into technical proposals and data requirements.
- Assess technical feasibility, define assumptions, acceptance criteria, and evaluation requirements.
- Create detailed rubrics, counterexamples, and boundary cases to define quality standards.
- Perform deep audits of produced data to identify subtle errors, reward hacking, leakage, and distribution shifts.
- Drive iterative improvements by analyzing failures and providing concrete instructions to the production team.
- Design and build datasets and RL environments for capability areas such as coding/SWE agents, multimodality, STEM reasoning, and embodied AI.
- Contribute to the design of task suites, ground-truth signals (verifiers, reward functions), and environment interfaces (APIs, tool schemas).
- Build robust validation, denoising, and synthetic data systems.
- Implement automated validation and filtering for signal-to-noise ratio, deduplication, decontamination, and consistency checks.
- Develop synthetic data generation and augmentation pipelines.
- Create documentation and data cards detailing dataset intent, limitations, and usage.
- Design and run evaluations to prove the impact of data and training runs.
- Produce analysis connecting data to outcomes, including pre/post comparisons and error breakdowns.
Requirements
- Experience building and improving deep learning systems.
- Experience with data quality, data curation, denoising, synthetic data generation, and rigorous evaluation.
- Experience in one or more of the following capability areas: coding and software engineering agents, RL environments and verifier-based training, multimodal data and reasoning, STEM reasoning, or modern embodied AI / VLM-driven agents.
- Ability to translate ambiguous research goals into technical proposals and data requirements.
- Ability to define quality standards through detailed rubrics and counterexamples.
- Ability to perform deep audits of data and identify subtle errors.
- Ability to drive iterative improvements based on evidence.
- Ability to design and build datasets and RL environments.
- Ability to build robust validation, denoising, and synthetic data systems.
- Ability to implement automated validation and filtering.
- Ability to develop synthetic data generation and augmentation pipelines.
- Ability to create documentation and data cards.
- Ability to design and run evaluations.
- Ability to produce analysis connecting data to outcomes.