Principal Applied Scientist (Agentic AI)

TL;DR

Principal Applied Scientist (Agentic AI): Leading the design and deployment of RL post-training systems to align large models with user value and safety with an accent on preference modeling and multi-objective optimization. Focus on developing reward models, implementing RLHF/DPO pipelines, and scaling AI-powered experiences within the real estate domain.

Location: Remote (USA). Must be based in the United States.

Salary: $181,800 – $305,700 annually

Company

hirify.global is the most-visited real estate platform in the U.S., helping customers navigate buying, selling, financing, and renting.

What you will do

• Lead the technical direction and strategy for RL post-training of production models.
• Design and implement post-training pipelines using SFT, DPO, RLHF, and RLAIF.
• Develop reward models and objective formulations balancing helpfulness, safety, and compliance.
• Translate conversational logs and behavioral signals into actionable supervision for reinforcement learning.
• Collaborate with platform teams to optimize training efficiency, off-policy evaluation, and rollout metrics.
• Mentor applied scientists and engineers to raise the technical bar in RL and evaluation.

Requirements

• PhD or equivalent experience in Computer Science, Electrical Engineering, Statistics, or a related field.
• Strong expertise in post-training techniques including SFT, DPO, RLHF, and preference modeling.
• Proficiency with transformer-based models, LLMs, multimodal models, and vector search.
• Experience in high-stakes domains where safety, trust, or regulation are critical (e.g., finance, healthcare).
• Proven technical leadership and mentorship experience.
• Must be based in the USA.

Culture & Benefits

• Remote-first work environment emphasizing experimentation, learning, and rapid shipping.
• Competitive base salary and eligibility for equity awards.
• Inclusive culture recognized by Fortune 100 Best Companies to Work For.
• Opportunity to represent company work through external talks, publications, and open-source contributions.