Job Description
We’re looking for an AI Research Engineer who blends frontier research curiosity with engineering discipline. You’ll join the Jan and Tokamak model team, training a state-of-the-art model that can carry out various jobs (including long running deep research tasks) for millions of users.
This role is ideal for someone who thrives in high-performance environments, understands the nuances of training LLMs/VLAs/etc, and is obsessed with fast experimentation and applied usability.
You are a great fit if you are
- Intellectually Curious
- Productively Combative
Responsibilities
- Obsess over model usability for users
- Conduct training runs and AI experiments
- Scrutinize results and iterate
- Continuously productionize checkpoints with the product engineering team
What you'll do
- Iterate upon latest RL techniques, e.g. GRPO, DPO, RePO, etc.
- Build and maintain modular, scalable training codebases
- Develop and maintain efficient data pipelines (synthetic and real)
- Ensure training jobs can scale across multiple GPUs and nodes (e.g., FSDP, DDP, NCCL)
- Maintain long-term code health: write clean, testable and reproducible code
- Contribute to upstream open source dependencies
- (Optionally) Publish papers and present findings
Requirements
- Deep expertise in Python and training codebases, e.g. PyTorch, or equivalent
- Proven experience training deep learning & reinforcement learning models in real-world settings
- Experience working with large datasets, complex pipelines
- Understanding of training dynamics: what goes wrong, and how to fix it
- Familiarity with job launchers, logging tools (e.g., Weights & Biases, TensorBoard), and checkpointing systems
- A mindset of engineering rigor applied to research: readable code, thoughtful design, and reproducibility
Bonus Points
- Experience with TorchScript, ONNX, or custom inference runtimes
- Open Source contributions to PyTorch or ML tooling
- Experience working on transformer models, diffusion models, VLMs, or large-scale vision/NLP tasks
- Familiarity with batch schedulers (SLURM), cluster environments, and GPU resource management
- Ability to collaborate closely with systems engineers or MLOps teams to ensure smooth integration
About Jan
Jan is an Personal AI Assistant that helps you get things done. It features a powerful local mode that keeps your data fully private.
Try Jan: https://jan.ai
Jan models: https://huggingface.co/janhq
Why: https://www.jan.ai/handbook/why/open-superintelligence
