AI Model Quality Engineer – Open-Weight Training Lead

Sapienceai; Seattle, Northern (WA, KY); Hybrid; USD 204,000 - 216,000; Full time

Benefits offered by this job

Health and wellness benefits

Job summary

Sapience AI is hiring an AI Model Training Quality Engineer to own the discipline of training, reinforcement, and evaluation for open‑weight foundation models. You will set the standard for model quality, lead end‑to‑end processes, and ensure safety, groundedness, and reliability as models are brought to production.

You will collaborate with Research, Neuro‑Symbolic AI, and ML Infrastructure to align training with community needs, while shaping data quality, evaluation frameworks, and production

Qualifications

• Five+ years of hands-on ML training and fine-tuning of modern models.
• Direct experience training or fine-tuning open-weight large language models, including supervised fine-tuning and domain adaptation.
• Hands-on experience with reinforcement and alignment methods (e.g., RLHF, RLAIF, or preference optimization).
• Deep experience designing and running model evaluation for accuracy, groundedness, calibration, safety.

Responsibilities

• Define and uphold the training-quality bar for Sapience AI's models.
• Lead training, reinforcement, and evaluation end-to-end to ship reliable models.
• Own data quality, evaluation datasets, and contamination control for training and testing.

Skills

ML training

Open-weight LLMs

Reinforcement learning

Evaluation design

Python

Distributed training

Tools

PyTorch

DeepSpeed

Megatron

Experiment tracking

Job description

Sapience AI is hiring an AI Model Training Quality Engineer to own the discipline of training, reinforcement, and evaluation for open‑weight foundation models. You will set the standard for model quality, lead end‑to‑end processes, and ensure safety, groundedness, and reliability as models are brought to production.

You will collaborate with Research, Neuro‑Symbolic AI, and ML Infrastructure to align training with community needs, while shaping data quality, evaluation frameworks, and production