Machine Learning Engineer

A stealth-mode startup is developing a groundbreaking product in the field of Foundation Models and Multimodal AI.

The company is based in Tel Aviv near the light rail, employs around 12 people locally, and operates a flexible hybrid model (2–3 days working from home). It also has a presence in the U.S. with a cross-continental research team.

The company is led by a rare team of AI researchers, all PhDs in computer science, and has already raised $12.5 million at the pre-seed stag

About the role-

• Join a strong team combining hands-on development with applied research
• Collaborate with the research team in Silicon Valley (local working hours)
• Develop and lead distributed training infrastructure on GPU-heavy clusters
• Apply advanced techniques such as ZeRO Optimization, Mixed Precision, Flash Attention, FP8 Training, Vision Transformers, CLIP, and Cross-Attention
• Build advanced pipelines for Image-Text datasets
• Work with PyTorch and JAX in multi-node environments with hundreds of GPUs
• Ensure high standards of technological excellence and competitiveness

Requirements-

• Minimum of a research-based master’s degree in CS, Mathematics, Physics, or related fields (PhD preferred)
• 3 years of industry experience in Machine Learning – required
• Hands-on experience with Deep Learning – required
• Experience training large-scale models (100B+ parameters) – required
• Experience building distributed training infrastructure using tools such as PyTorch/JAX, ZeRO Optimization, or Mixed Precision – required
• Understanding of Transformer architectures – strong advantage
• Experience building pipelines for Image-Text datasets – advantage
• Scientific publications on models with 10B+ parameters – advantage
• Experience training on 100+ GPUs – advantage
• Previous experience training models at leading global companies such as Meta, Google, AI21, or AI Labs – strong advantage

150862