🛠 Tech Stack

💼 About This Role

You'll build end-to-end ML pipelines and own the execution layer of A1's intelligence. Your work will power production-grade inference systems serving billions globally. This role combines research-to-production engineering with real ownership of latency, cost, and safety.

🎯 What You'll Do

• Build and own end-to-end ML training and inference pipelines
• Fine-tune models using LoRA, SFT, DPO, and distillation
• Design scalable inference systems balancing latency and cost
• Implement evaluation pipelines for performance and safety
• Own production deployment with GPU optimization

📋 Requirements

• Deep learning and transformer-based architectures background
• Hands-on production ML training or deployment experience
• Proficiency with PyTorch or JAX
• Experience with distributed training frameworks like DeepSpeed or FSDP
• Strong software engineering fundamentals

✨ Nice to Have

• Experience with LLM inference frameworks like vLLM or TensorRT-LLM
• Contributions to open-source ML libraries
• Background in GPU kernels or scientific computing

🎁 Benefits & Perks

• 🚀 High talent density team with rapid execution
• 🌍 Remote work flexibility
• 🧠 Ownership of zero-to-one AI systems
• 🏆 Impact on a product serving billions

📨 Hiring Process

3-4 interviews including technical team evaluation, conducted virtually or onsite, with prompt decision.