Hybrid ML Engineer, Safety & AI Agents

Lyft; Toronto; Hybrid; CAD 119,000 - 149,000; Full time

Benefits offered by this job

Extended health and dental coverage

Mental health benefits

Family building benefits

RRSP plan with company match

Paid time off (flexible)

Job summary

Lyft is seeking a Machine Learning Engineer for the Safety and Customer Care (SCC) team in Toronto on a hybrid schedule. You will fine-tune models, build AI agents, and design evaluation loops for safety-focused support systems.

The role focuses on post-training with open-source models, RLHF/RLAIF, and agentic development in a production setting. Expect collaboration with ML, data science, backend, and product teams to scale agents for millions of interactions while maintaining safety and

Qualifications

• 3+ years of industry experience in applied ML/AI with an MS or PhD.
• Post-training experience with open-source models and fine-tuning methods.
• Agentic development experience with LangGraph or equivalent frameworks.
• Strong Python programming and PyTorch experience.
• Publications at top-tier venues preferred.

Responsibilities

• Conduct literature review and build post-training framework and lifecycle.
• Curate and process human and synthetic data for SFT/LoRA/RLHF/RLAIF/RLVR.
• Develop, evaluate, and productionize AI agents with LangGraph (or equivalent).
• Build and scale evaluation frameworks, rubrics, and LLM-as-judge methods.
• Ship models into real-time production with monitoring and guardrails.
• Collaborate with product, ops, and data science to define success metrics.

Skills

Python

PyTorch

Agentic development

LangGraph

Open-source models

RLHF/RLAIF/RLVR

Education

MS or PhD in Computer Science, ML, AI or related field

Tools

LangGraph

Post-training frameworks

Job description

Lyft is seeking a Machine Learning Engineer for the Safety and Customer Care (SCC) team in Toronto on a hybrid schedule. You will fine-tune models, build AI agents, and design evaluation loops for safety-focused support systems.

The role focuses on post-training with open-source models, RLHF/RLAIF, and agentic development in a production setting. Expect collaboration with ML, data science, backend, and product teams to scale agents for millions of interactions while maintaining safety and