Fine-tune & adapt - Use LoRA/QLoRA to optimize open-source models (LLaMA, Mistral, Gemma)
Engineer prompts & curates data - Craft prompts and datasets that reflect tone, brand voice, and safety.
Evaluate models – Build metrics pipelines for perplexity, toxicity, and relevance to ensure safe and high-quality outputs.
Deploy & monitor – Scale models into production with performance optimization and monitoring for drift.
Collaborate & deliver – Partner with product, engineering, and design teams to launch user-facing AI features.
You’ll fine-tune state-of-the-art models, design evaluation frameworks, and bring AI features into production. Your work ensures our models are not only intelligent, but also safe, trustworthy, and impactful at scale.
Requirements
Strong experience in transformers, deep learning, and fine-tuning methods (LoRA/QLoRA, SFT, distillation).
Proficiency with PyTorch (preferred) or TensorFlow.
Skilled in prompt engineering and dataset curation for alignment with tone, safety, and trust.
Familiar with evaluation metrics: perplexity, toxicity, relevance.
Strong software engineering foundations in algorithms, data structures, and clean code practices.