Deskripsi Pekerjaan
At Nexus Intelligence Labs, we are architecting the next generation of autonomous AI agents that redefine how enterprises interact with data. We are seeking a Senior Machine Learning Engineer to lead the design, training, and deployment of large-scale foundation models in a production environment.
You will operate at the intersection of research and engineering, collaborating with a world-class team of researchers and infrastructure engineers to push the boundaries of LLM fine-tuning, RAG architectures, and multimodal model serving. This is a high-impact role where your code ships to millions of end-users daily.
Why Join Us?
- Ownership: End-to-end ownership of model lifecycle—from data curation to production monitoring.
- Compute Budget: Unrestricted access to H100/A100 clusters for experimentation and training.
- Culture: Low bureaucracy, high trust, 'write the code, run the experiment' mentality.
- Compensation: Top-of-market salary, significant early-stage equity, 401k matching, and comprehensive health/wellness stipends.
Tanggung Jawab
- Design, implement, and optimize distributed training pipelines for LLMs (7B–70B+ parameters) using PyTorch, FSDP/DeepSpeed, and Megatron-LM.
- Architect and maintain low-latency, high-throughput model serving infrastructure (vLLM, TGI, TensorRT-LLM) on Kubernetes (EKS/GKE).
- Lead the development of robust evaluation frameworks (automated + human-in-the-loop) for alignment, hallucination detection, and safety guardrails.
- Spearhead RAG and Agentic workflow optimization: chunking strategies, advanced retrieval (HyDE, rerankers), and tool-use reliability.
- Collaborate with Product and Research to translate cutting-edge papers (ICLR, NeurIPS, ICML) into production-grade features within weeks.
- Establish MLOps best practices: CI/CD for models, feature stores, data lineage, automated drift detection, and cost-optimized inference scaling.
- Mentor junior engineers and conduct rigorous code/design reviews ensuring system reliability and maintainability.
Kualifikasi
- 5+ years of professional Software Engineering experience with 3+ years specializing in ML Infrastructure / MLOps / GenAI.
- Expert-level proficiency in Python (type-hinted, async, testing) and modern C++ (for kernel optimization/custom ops).
- Deep hands-on experience with Distributed Training (DDP, FSDP, DeepSpeed, Megatron) on multi-node GPU clusters.
- Proven track record of optimizing LLM inference: quantization (GPTQ, AWQ, bitsandbytes), speculative decoding, KV-cache management, and batching strategies.
- Strong systems background: Linux internals, networking (RDMA/RoCE), containerization (Docker/K8s), and GPU scheduling (Slurm/KubeRay).
- Experience with Data-centric AI: building petabyte-scale data pipelines (Spark/Ray Data), synthetic data generation, and curriculum learning.
- Publications at top-tier venues (NeurIPS, ICML, ICLR, MLSys) or significant open-source contributions to ML frameworks (vLLM, HuggingFace, PyTorch Lightning) highly preferred.