Deskripsi Pekerjaan
Shape the Future of Generative AI
At Nexus Intelligence Labs, we are building the foundational models powering the next generation of enterprise automation. We are looking for a Senior Machine Learning Engineer to join our core research team in the heart of San Francisco. This is a high-impact role where you will own the full lifecycle of large-scale model development—from data curation and architecture design to distributed training and production deployment.
Why This Role is Different
- ✦ Compute Access: Dedicated access to 4,000+ H100 GPUs via our strategic cloud partnerships.
- ✦ Research Freedom: 20% protected time for independent exploration and publication at top venues (NeurIPS, ICML, ICLR).
- ✦ Real-World Impact: Models deployed to Fortune 500 clients within weeks, not years.
We value first-principles thinking, clean code architecture, and a bias toward action. If you have a track record of shipping LLMs or diffusion models at scale, we want to talk.
Tanggung Jawab
- Design, implement, and optimize novel transformer architectures for multimodal foundation models.
- Lead distributed training runs across 100+ node clusters (PyTorch FSDP, Megatron-LM, DeepSpeed), maximizing MFU (Model FLOPs Utilization).
- Build and maintain high-throughput data pipelines (Petabyte-scale) using Spark/Ray for pre-training, instruction tuning, and RLHF.
- Collaborate with Research Scientists to translate theoretical breakthroughs into production-grade, reproducible code.
- Develop custom CUDA/Triton kernels to accelerate attention mechanisms and quantization workflows (INT4/GPTQ/AWQ).
- Establish rigorous evaluation frameworks (automated benchmarks + human preference alignment) to guide model iteration.
- Mentor junior engineers and contribute to the team’s technical strategy and hiring bar.
- Publish findings at top-tier conferences and open-source tooling to contribute to the community.
Kualifikasi
- Education: MSc or PhD in Computer Science, Machine Learning, or equivalent practical experience.
- Experience: 5+ years industry experience; 3+ years specifically training large-scale deep learning models (>1B parameters).
- Framework Mastery: Expert-level PyTorch internals (autograd, distributed, compile); strong familiarity with JAX or TensorFlow.
- Distributed Systems: Proven track record debugging and optimizing training across multi-node GPU clusters (InfiniBand/RoCE).
- Model Architecture: Deep understanding of Transformer variants, attention mechanics, positional encodings, and scaling laws.
- Optimization: Hands-on experience with mixed precision (FP8/BF16), activation checkpointing, gradient accumulation, and ZeRO stages.
- Software Engineering: Production-grade Python (type hints, testing, CI/CD), Rust/C++ for kernel work is a major plus.
- Communication: Ability to distill complex technical trade-offs for cross-functional stakeholders.