Turing is a startup building fully autonomous driving systems powered by an end-to-end AI stack and a large-scale foundation model.
<div class="content-intro"><h1><span style="font-family: helvetica, arial, sans-serif;"><strong><span style="font-size: 24pt;">About Turing</span></strong></span></h1> <p><span style="font-size: 10pt;">Based in San Francisco, California, Turing is the world’s leading research accelerator for frontier AI labs and a trusted partner for global enterprises looking to deploy advanced AI systems. Turing accelerates frontier research with high-quality data, specialized talent, and training pipelines that advance thinking, reasoning, coding, multimodality, and STEM. For enterprises, Turing builds proprietary intelligence systems that integrate AI into mission-critical workflows, unlock transformative outcomes, and drive lasting competitive advantage.</span></p> <p><span style="font-size: 10pt;">Recognized by Forbes, The Information, and Fast Company among the world’s top innovators, Turing’s leadership team includes AI technologists from Meta, Google, Microsoft, Apple, Amazon, McKinsey, Bain, Stanford, Caltech, and MIT. Learn more at<a class="c-link c-link--underline" href="http://www.turing.com/" target="_blank" data-stringify-link="http://www.turing.com" data-sk="tooltip_parent"> www.turing.com</a></span></p></div><p><strong><span style="font-size: 12pt;"><em>*This is a remote role and can be performed anywhere in Colombia.*</em></span></strong></p> <h2><strong>The Role</strong></h2> <p>We are looking for a Research Engineer to help deliver frontier-quality datasets, RL environments, and evaluations that improve state-of-the-art models for leading AI labs and enterprise clients.</p> <p>This is a hands-on, research-facing technical leadership role. You will work directly with customer researchers & engineers to translate their model and post-training goals into concrete data and environment specifications, and drive the production of data that meets extremely high standards for correctness, realism, diversity, difficulty, and measurable model lift.</p> <p>This role is designed for candidates with roughly 4 to 5 years of experience building and improving deep learning systems, especially where strong results depend on data quality, data curation, denoising, synthetic data generation, and rigorous evaluation. You’ll operate in one or more of the following capability areas:</p> <ul> <li><strong>Coding and software engineering agents</strong> (repositories, unit tests, debugging, tool use, code reviews, long-horizon workflows)</li> <li><strong>RL environments and verifier-based training</strong> (tasks, rewards/verifiers, trajectories, evaluation harnesses)</li> <li><strong>Multimodal data and reasoning</strong> (text + images + documents + tables/charts; optional audio/video)</li> <li><strong>STEM reasoning</strong> (math, physics, chemistry, bio, engineering – solution verification and error analysis)</li> <li><strong>Modern embodied AI / VLM-driven agents</strong> (vision-language(-action) models, embodied task suites, tool/sensor/action abstractions, long-horizon interaction data)</li> </ul> <h2><strong>What You’ll Do</strong></h2> <h4><strong>1) Own data and environment quality from an AI researcher perspective</strong></h4> <ul> <li>Translate ambiguous research goals into clear data requirements: target skills, failure modes, difficulty calibration, coverage, and success metrics.</li> <li>Define what “good” looks like by creating detailed rubrics, counterexamples, and boundary cases (what to include vs. exclude).</li> <li>Perform deep, detail-oriented audits of produced data: spot subtle errors, reward hacking opportunities, leakage, ambiguity, inconsistent assumptions, and distribution shifts.</li> <li>Drive iterative improvements using evidence: error taxonomies, slice-based quality metrics, and model-behavior-informed refinements.</li> </ul> <h4><strong>2) Design and build datasets and RL environments for your capability area(s)</strong></h4> <ul> <li>Contribute to or lead the design of:</li> <ul> <li><strong>Task suites</strong> (single-step and long-horizon workflows)</li> <li><strong>Ground-truth signals</strong> (verifiers, unit tests, structured checks, reward functions, automatic validators)</li> <li><strong>Environment interfaces</strong> (APIs, tool schemas, state abstractions, database schemas, simulator-like dynamics)</li> </ul> <li>Depending on your mapped capability area(s), you may focus on:</li> <ul> <li><strong>Coding / SWE agents:</strong> data reflecting real development work (codebase navigation, bug localization, patching, tests, code reviews, CI-like constraints, refactors, security fixes).</li> <li><strong>Multimodality:</strong> tasks that test true multimodal reasoning (chart reading, document QA, UI understanding, diagram-based STEM reasoning, OCR-aware tasks).</li> <li><strong>STEM:</strong> tasks with verifiable solutions (symbolic checks, reference solvers, numerical validation, step consistency, unit sanity).</li> <li><strong>Modern embodied AI / VLM-driven agents:</strong> interaction data and environments for vision-language(-action) models (long-horizon tasks, instruction following grounded in visual context, robust action selection, safety/constraint adherence, adversarial state coverage).</li> </ul> </ul> <h4><strong>3) Build robust validation, denoising, and synthetic data systems</strong></h4> <ul> <li>Implement automated validation and filtering to achieve frontier-grade signal-to-noise:</li> <ul> <li>Deduplication, decontamination, leakage checks</li> <li>Consistency checks (format, schema, invariants)</li> <li>Difficulty and diversity controls (coverage, novelty, long-tail)</li> </ul> <li>Develop synthetic data generation and augmentation pipelines where appropriate:</li> <ul> <li>Programmatic task generators</li> <li>Controlled perturbations to create hard negatives</li> <li>Scenario templating with diversity constraints</li> <li>Simulator-/tool-driven rollouts for trajectory data</li> </ul> <li>Create documentation and data cards: dataset in