Research Engineer, Mid-Training
San FranciscoFull-timePosted 6mo ago
About the role
Applied AI lab building end-to-end software agents. Developer of Devin, the first AI software engineer. Role focuses on late-stage training decisions to sharpen model capabilities across data, schedules, and synthetic data strategies.
Responsibilities
- Data Mix and Quality Uplift: Design and iterate on high-quality data mixtures for late-stage and annealing training runs. Develop principled methods for sourcing, filtering, and weighting data to sharpen model capabilities without degrading general performance.
- Capability Injection: Drive targeted improvements in coding, mathematics, and long-horizon reasoning through curated data strategies and training interventions. Translate research insights into measurable capability gains on our agents.
- Synthetic Data Research: Develop and evaluate synthetic data pipelines that generate training signal at scale. Understand the limits and failure modes of synthetic approaches and build methods that hold up in production training runs.
- Annealing and Schedule Design: Research and optimize multi-stage learning rate schedules, warmup strategies, and compute allocation across training phases. Understand how schedule choices interact with data distribution and model behavior.
- Context Length Extension: Research and implement methods for extending effective context length without degrading short-context performance, including positional encoding strategies and data construction.
- Evaluation and Iteration: Build evals that distinguish real capability improvements from benchmark overfitting. Close the loop between training decisions and deployed systems.
- Scaling and Methodology: Measure how mid-training interventions scale with compute and data; develop new approaches when existing methods hit ceilings.
Qualifications
- Deep familiarity with the LLM training pipeline end to end: pre-training data, optimization, architecture, and mid-training and post-training interactions
- Hands-on experience with continual pre-training, annealing, or late-stage data mixing for large models
- Strong intuition for data quality: dataset usefulness, scalable filtering and curation, data mix impact across evals
- Experience developing or evaluating synthetic data pipelines for capability improvement
- Proficiency in Python and deep learning frameworks (PyTorch); comfortable debugging distributed training at scale
- Strong fundamentals in optimization, statistics, and ML theory; ability to distinguish real effects from noise
- Track record of original contributions: publications, open-source impact, or internal results that moved a capability frontier
- Comfort operating in ambiguous, fast-moving environments
- PhD is welcome but not required; not a sole criterion
Skills
Required:LLM training pipelineContinual pre-training / annealing / late-stage data mixingData quality and curationSynthetic data pipelines
Benefits
- Unspecified in description, not included
Last updated Oct 21, 2026. Interested? Apply on the company site.
About Cognition
cognition is an applied AI lab
An applied AI lab building the future of software engineering. We’re the makers of Devin, a collaborative AI teammate that helps ambitious engineering teams achieve more.
AIai-plg-top-20collaborative AIliquid-ai-recruiting-targetmachine learningsoftware engineering
More open roles at Cognition
- Sales Development Manager4d agoSan Francisco, Austin, New York City · Sales
- Partner Director (LATAM)4d agoSao Paulo · Sales
- Analyst Relations4d agoSan Francisco · Marketing
- Head of Safety1w agoSan Francisco · Research & Development
- Japan Partner Marketing1w agoTokyo · Marketing