Deskripsi Pekerjaan
Nexus Intelligence is at the forefront of applying Large Language Models to enterprise knowledge management. We are seeking a Senior Machine Learning Engineer to architect and scale our core inference platform. This is a high-autonomy role for a builder who thrives in a fast-paced, research-adjacent environment.
You will own the end-to-end lifecycle of production models, from data curation and training optimization to low-latency serving infrastructure. We value clean code, rigorous evaluation, and a product-first mindset.
- Hybrid Model: 3 days/week in our SoMa HQ (555 California St).
- Compensation: Top-of-market base + significant pre-IPO equity.
- Stack: Python, PyTorch, JAX, Triton, Kubernetes, AWS, Ray.
Tanggung Jawab
- Design, train, and optimize foundation models and LLMs for specific enterprise use cases (RAG, Agents, Structured Extraction).
- Build and maintain scalable MLOps pipelines for data processing, training, evaluation, and deployment (CI/CD for ML).
- Optimize model inference latency and throughput using techniques like quantization, distillation, and speculative decoding.
- Collaborate with Product and Research to define model requirements and evaluation benchmarks.
- Architect robust, observable model serving infrastructure on Kubernetes (EKS) using Triton/TorchServe.
- Conduct rigorous offline/online evaluation (A/B testing, shadow traffic) to validate model improvements.
- Mentor junior engineers and establish best practices for code quality, reproducibility, and security.
- Stay current with SOTA literature (e.g., Mixture of Experts, Long Context, Multimodality) and prototype promising architectures.
Kualifikasi
- 5+ years industry experience in Machine Learning Engineering; 2+ years deploying LLMs/Generative AI in production.
- Expert proficiency in Python and Deep Learning frameworks (PyTorch required; JAX/TensorFlow a plus).
- Deep understanding of Transformer architecture, attention mechanisms, and distributed training (FSDP, DeepSpeed, Megatron-LM).
- Proven track record optimizing GPU utilization and inference serving stacks (vLLM, TGI, Triton, TensorRT-LLM).
- Strong software engineering fundamentals: Linux, Docker, Kubernetes, CI/CD, Git, System Design.
- Experience with Cloud platforms (AWS preferred: SageMaker, EC2, S3, EFS) and Infrastructure as Code (Terraform).
- M.Sc. or Ph.D. in Computer Science, Machine Learning, or equivalent quantitative field.
- Publications at top-tier venues (NeurIPS, ICML, ICLR, ACL) or significant open-source contributions preferred.