Applied Machine Learning Researcher

About the role

Phonely is building conversational voice‑AI agents for high‑volume phone workflows. As an Applied Machine Learning Researcher you will improve the naturalness, reliability, and speed of our voice AI models, delivering real impact in production.

Key responsibilities

• Research and improve LLM behavior for real‑time voice conversations
• Design and run fine‑tuning experiments across data, model, and evaluation strategies
• Build evaluation frameworks for model quality, workflow‑following, naturalness, reliability, task completion, and overall conversation quality
• Analyze production conversations to find failure modes and opportunities to improve
• Develop data curation, labeling, and synthetic data strategies
• Compare model architectures, training approaches, prompts, and datasets
• Investigate regressions and explain why model behavior improves or degrades
• Work with engineering to deploy research improvements safely and efficiently
• Help define model release criteria, evaluation gates, and quality benchmarks

Required profile

• Strong experience with machine learning and modern language models
• Hands‑on experience fine‑tuning, evaluating, or adapting language models
• Strong Python skills and comfort working with messy, real‑world datasets
• Ability to design rigorous experiments and interpret results clearly
• Experience building or improving evaluation systems for AI models
• Strong analytical skills and ability to debug model behavior
• Clear written and spoken communication for technical and non‑technical audiences
• Practical mindset focused on production impact, latency, reliability, and customer outcomes

Required skills

• Python
• Machine learning
• Large language models (LLMs)
• Fine‑tuning techniques
• Evaluation framework development
• Data curation and synthetic data generation