Build the next generation of foundation-model capabilities at Mecka AI using Python, PyTorch, large-scale experiments, and advanced post-training.

This experienced Research Scientist will work with large-scale datasets and distributed training infrastructure. You will train, fine-tune, and evaluate multimodal models, including Qwen or comparable open-weight systems. The role emphasizes practical experimentation, data quality, reinforcement learning, and measurable model improvements.

Key Responsibilities:

- Train and post-train large-scale multimodal foundation models
- Design data curation, filtering, and synthetic-data generation strategies
- Execute and analyze large-scale machine learning experiments
- Investigate model optimization and performance improvements




- Explore reward modeling and advanced model evaluation

Requirements:
- Experience with Qwen, Qwen-VL, or comparable open-weight models
- Proficiency in Python and PyTorch, JAX, or similar frameworks
- Experience with supervised fine-tuning or reinforcement learning
- Familiarity with multimodal images, video, or sensory-data pipelines
- Publications, open-source contributions, or leading AI organization experience

Use foundation-model training, data strategy, and experiment analysis skills to advance multimodal AI research at Mecka AI.

#J-18808-Ljbffr

📌 Foundation Model Research Scientist (Canada)
🏢 mecka
📍 Canada