Deskripsi Pekerjaan
At Nexus Intelligence Labs, we are architecting the future of generative AI for enterprise automation. We are looking for a Senior Machine Learning Engineer to lead the development of our core Large Language Model (LLM) infrastructure, optimizing model training pipelines and deploying scalable inference systems that power our flagship products.
You will sit at the intersection of research and production, working closely with Research Scientists to bring novel architectures to life and Platform Engineers to ensure reliability at scale. This is a high-impact role where your work directly shapes the intelligence layer of products used by Fortune 500 companies globally.
We offer a culture of technical excellence, compute budgets that remove bottlenecks, and the autonomy to explore state-of-the-art techniques—from Mixture of Experts (MoE) routing to advanced quantization and speculative decoding.
Tanggung Jawab
- Design, build, and maintain high-throughput, low-latency ML training and inference pipelines on GPU clusters (Kubernetes/Slurm).
- Lead the optimization of LLM serving stacks (vLLM, TensorRT-LLM, TGI) for cost-efficiency and latency SLOs.
- Collaborate with Research Scientists to productionize novel model architectures (MoE, RAG, Agentic workflows).
- Implement robust MLOps practices: experiment tracking, model versioning, automated evaluation suites, and CI/CD for ML.
- Develop custom kernels or optimization strategies (FlashAttention, quantization-aware training) for proprietary hardware accelerators.
- Mentor junior engineers and define technical standards for code quality, testing, and documentation.
- Participate in on-call rotations for critical production inference endpoints.
Kualifikasi
- 5+ years of professional software engineering experience, with 3+ years focused on ML Infrastructure/Platform teams.
- Expert proficiency in Python and PyTorch/JAX; strong systems programming skills in C++ or Rust.
- Deep hands-on experience with distributed training (FSDP, DeepSpeed, Megatron-LM) and large-scale inference optimization.
- Proven track record deploying and operating LLMs (7B–70B+ params) in production environments.
- Strong expertise with container orchestration (Kubernetes, Docker) and cloud infrastructure (AWS/GCP/Azure), specifically GPU node management.
- Solid understanding of GPU architecture (H100/A100), CUDA programming basics, and memory hierarchy optimization.
- Experience with data processing frameworks (Spark, Ray, Polars) for petabyte-scale dataset curation.
- Excellent communication skills; ability to articulate complex technical trade-offs to cross-functional stakeholders.