Staff Maching Learning Engineer - Fine - tuning

Machine Learning Engineer

Snowflake is about empowering enterprises to achieve their full potential - and people too. With a culture that's all in on impact, innovation, and collaboration, Snowflake is the sweet spot for building big, moving fast, and taking technology - and careers - to the next level. The Snowflake Machine Learning Platform team's mission is to enable customers to bring their ML/AI workloads to Snowflake. Our customers want to leverage ML/AI to extract business value from ever-increasing data in Snowflake but face challenges across infrastructure management, scaling, orchestration, performance, and security. The team solves these challenges by building deeply integrated platform solutions that are simple, secure, and scalable to enable end-to-end ML workflows. We're early in our journey to build the best machine learning and data platform for Snowflake customers while preserving the benefits of a single, governed platform.

As a ML Engineer focused on fine-tuning, you'll own the technical vision and delivery of customer-facing training workflows and APIs - from data prep to evaluation and rollout. You'll translate the latest research (across SFT, RFT, prompt tuning) into reliable, scalable systems that work at enterprise scale and within Snowflake's security and governance model to deliver direct business value.

YOU WILL:

• Define and build an end-to-end fine-tuning platform (APIs, SDKs, and UI) that makes it easy for customers to adapt foundation models on their own data inside Snowflake, with first-class governance, privacy, and lineage.
• Collaborate closely with Product and Snowflake Research; stay current with the state of the art and convert research into cohesive, production-ready systems and product capabilities.
• Partner with strategic customers to deeply understand pain points, design delightful workflows, run pilots, and iterate quickly based on data-driven feedback.
• Implement distributed training at scale to maximize user value and minimize cost and end-to-end latency
• Build robust evaluation and safety tooling (automatic and human-in-the-loop)

QUALIFICATIONS:

• 7+ years building ML systems end-to-end with shipped, production systems; proven ability to lead initiatives with a product mindset.
• Expertise in distributed training and systems performance (e.g., PyTorch, FSDP/DeepSpeed, CUDA/NCCL, Ray/Kubernetes); strong debugging skills across data, modeling, and infrastructure layers.
• Deep working knowledge of LLMs and fine-tuning methods: SFT, DPO/RFT, RLHF fundamentals; parameter-efficient tuning (LoRA/QLoRA, adapters, prompt tuning)
• Experience building evaluations for capability improvement and safety (task-level benchmarks, offline/online metrics, A/B testing, human review pipelines, guardrails).
• Track record of implementing and hardening complex ML systems with rigorous reliability, latency/throughput optimization, and cost efficiency.
• BS in Computer Science or related field (MS/PhD preferred).

WHY JOIN OUR TEAM AT SNOWFLAKE? You'll be at the forefront of innovation in machine learning infrastructure, working with cutting-edge technologies and solving complex challenges. Our team is driving the future of...