Deskripsi Pekerjaan
At Nexus Intelligence Labs, we are architecting the next generation of autonomous AI agents for enterprise workflow automation. We are looking for a Senior Machine Learning Engineer to lead the design, training, and deployment of large-scale foundation models and retrieval-augmented generation (RAG) systems.
You will operate at the intersection of research and production, owning the full ML lifecycle from data curation and distributed training on GPU clusters to low-latency serving infrastructure. This is a high-impact role with direct ownership over core intellectual property.
Why Join Us?
- Work on frontier LLMs (70B+ parameters) with massive proprietary datasets.
- Compute budget is unlimited (dedicated H100/A100 clusters).
- Publish at top-tier venues (NeurIPS, ICML, ICLR) encouraged.
- Competitive equity package (ISO options) with 409A valuation refresh annually.
Tanggung Jawab
- Design and implement scalable model architectures (Transformers, MoE, RAG pipelines) optimized for throughput and latency.
- Lead distributed training runs (FSDP, DeepSpeed, Megatron-LM) across 100+ GPU nodes.
- Build and maintain robust MLOps pipelines (Kubeflow, MLflow, Airflow) for CI/CD/CT (Continuous Training).
- Develop advanced evaluation frameworks (human-in-the-loop, LLM-as-a-judge, red-teaming) for alignment and safety.
- Collaborate with Product and Data Engineering to define data flywheels: active learning, synthetic data generation, preference optimization (DPO/PPO).
- Optimize inference serving (vLLM, TensorRT-LLM, Triton) for sub-100ms p99 latency at scale.
- Mentor junior engineers and establish best practices for code quality, reproducibility, and experiment tracking.
- Stay current with SOTA literature; prototype and productionize novel techniques (e.g., speculative decoding, quantization-aware training).
Kualifikasi
- 5+ years industry experience in ML Engineering; 2+ years specifically on Large Language Models or Generative AI.
- Expert proficiency in Python, PyTorch/JAX, CUDA; deep understanding of GPU memory management and kernel fusion.
- Proven track record of training/fine-tuning models >7B params on distributed infrastructure (Slurm, Kubernetes, Ray).
- Strong software engineering fundamentals: CI/CD, Docker, Kubernetes, Terraform, GitOps.
- Deep knowledge of Transformer internals: Attention mechanisms, RoPE, FlashAttention, Quantization (GPTQ, AWQ, GGUF).
- Experience with RAG systems: Vector DBs (Milvus, Weaviate, Pinecone), Embedding models, Re-ranking, Agentic workflows (LangGraph, AutoGen).
- Publications at top-tier conferences (NeurIPS, ICML, ICLR, ACL, EMNLP) or significant open-source contributions (HuggingFace, vLLM, PyTorch).
- MS/PhD in Computer Science, Machine Learning, Statistics, or equivalent practical experience.