Research Engineer (Agentic Models)
Core
Designing, training, and evaluating multi-step coding agents that understand codebases, plan changes, and interact with users within JetBrains IDEs.
Role type
Senior Research Engineer (Agentic Models)
Builds
Multi-step coding agents, SFT/RL post-training pipelines, and evaluation environments for JetBrains products.
Domain
Developer tools / AI Agents / Large Language Models
Deliverable
production ML models
Required skills
LLM pre-training/fine-tuning/post-training, PyTorch, distributed GPU training, data pipeline design, agent workflow design, Python, model architecture design, training recipe optimization
Preferred skills
ML orchestrators (Kubeflow, Dagster, Airflow), large-scale data processing (MapReduce), experiment tracking (W&B, MLflow), inference optimization, agentic frameworks
Technologies
PyTorch, Megatron, NeMo, verl, Kubernetes, SLURM, Kubeflow, Dagster, Airflow, ZenML, W&B, MLflow, Langfuse
Responsibilities
Design and maintain SFT and RL post-training pipelines for multi-step coding agents; Train and adapt LLMs for agent workflows including planning and tool use; Build evaluation and simulation environments for coding agents; Design evaluation frameworks and metrics for agent behavior; Analyze training results to improve model architectures and datasets; Work with large-scale distributed GPU clusters and MapReduce-style data processing
Seniority
Senior, hands-on IC
