Microsoft

This role focuses on the co-design and productionization of next-generation AI systems at datacenter scale, optimizing performance, efficiency, and cost across hardware and software. It involves analyzing workloads, driving architectural decisions, optimizing distributed systems for training and inference, and influencing AI hardware design. The role also includes performance modeling, prototyping, and mentoring.

Microsoft AI is seeking individuals to train the world's most capable AI frontier models, focusing on scale, performance, and product deployment. The role involves developing algorithms, model architectures, and data mixtures for large-scale training, driving implementations, conducting experiments, and collaborating with infrastructure, data, post-training, and multimodality teams.

Research role focused on developing and training world-class multimodal AI frontier models, pushing boundaries of scale and performance. Involves algorithm development, model architecture design, experimentation, data innovation, and improving training/deployment efficiency. Collaborates with infrastructure, data engineering, pre-training, post-training, and product feedback teams. Requires strong publication track record and expertise in multimodal research.

Develops and ships post-training methods for large language models (LLMs) used in Microsoft Copilot, focusing on data collection, evaluation, finetuning algorithms, and prototyping new capabilities. Collaborates with pretraining and product platform teams to deploy models and improve them based on user feedback.

This role focuses on building and optimizing training data pipelines and evaluation frameworks for audio models, with a secondary focus on model tuning and deployment. The engineer will be responsible for data sourcing, cleaning, labeling, QA, versioning, and lineage, as well as developing evaluation frameworks and driving continuous quality improvements. They will also optimize training and inference performance for latency, throughput, cost, and reliability.

The role focuses on integrating custom AI silicon with AI inference frameworks like SGLang, optimizing LLM inference performance, and developing custom operators. It involves working with hardware accelerators and potentially non-CUDA ecosystems, aiming to improve AI workload efficiency.

This role focuses on designing, scaling, and optimizing high-performance networks for AI training and inference clusters. The engineer will work on the end-to-end networking architecture, from link-layer to fabric-wide systems, connecting thousands of GPUs. Responsibilities include benchmarking, profiling, debugging, and tuning AI workloads, engineering ultra-low-latency networks, and designing congestion-free transport mechanisms. The goal is to build networking systems that directly accelerate Microsoft's frontier AI models and support the development of advanced AI systems.

Principal Program Manager for Strategic Innovation within Microsoft Security Office of the CTO, focusing on identifying, shaping, and driving portfolio-level security incubations for Security for AI and GenAI-driven defense. The role involves translating emerging AI-driven risks into executable strategies and roadmaps, collaborating with product and engineering teams, and engaging with Fortune 500 CISOs.

This role is focused on ensuring the reliability, performance, and availability of Microsoft's large-scale AI training infrastructures, which involve tens of thousands of GPUs and advanced networking. The responsibilities include designing transport, fabric architecture, telemetry, observability, and automated troubleshooting for these clusters. The role also involves AI training and inference cluster bring-up, performance benchmarking, and root-cause analysis, with a goal of developing predictive health models and autonomous remediation systems.

Seeking Data Research Engineers to join the Multimodal team, focusing on building next-generation foundation models. The role involves exploring, designing, and building high-quality multimodal datasets (vision, language, audio) for training and evaluation, collaborating with scientists and engineers, and developing scalable data pipelines. Responsibilities include data analysis, quality assessment, building tools for auditing, and ensuring datasets meet responsible AI practices.