Research Engineer

Job Summary

The Research Engineer is pivotal in transforming how machines understand and replicate human conversation. The role focuses on building robust measurement frameworks, developing innovative data sourcing strategies, and fine-tuning algorithms that enhance machine conversation dynamics. By identifying flaws in existing systems and implementing effective solutions, this position aims to significantly close the gap between human and machine interactions.

Responsibilities

• Analyze real voice calls to identify failures in conversational dynamics, including interruptions, backchannels, and end-of-turn detection.
• Map the entire architecture of voice agents, ensuring that every component is accurately associated with voice-quality evaluations.
• Investigate inaccuracies within evaluation data, troubleshooting components for better accuracy in turn-taking detection.
• Develop innovative strategies for sourcing relevant data to optimize model training, harnessing external resources such as podcasts and video conversations.
• Fine-tune machine learning models and assess their performance within the live pipeline, ensuring a feedback loop exists between evaluation metrics and model performance.
• Research and prioritize open problems related to voice interaction, contributing to the design and implementation of new solutions.

Qualifications

• Advanced understanding of voice technology and machine learning, with proven experience in model fine-tuning and latency optimization.
• Experience in publishing research, showcasing effective strategies in the area of voice conversation technology.
• Proficient in working with complex systems, such as robotics or open-source projects, contributing meaningfully to large-scale architectures.
• Strong analytical skills with the ability to derive insights from data and research findings.
• A relevant degree in Computer Science, Engineering, or a related field; advanced degrees (Master's or PhD) preferred.

Preferred Skills

• Familiarity with Hugging Face and demonstrated success in deploying fine-tuned models for real-time applications.
• Background in speech-to-text (STT) technologies and bilingual/multilingual conversational systems.
• Experience in research that includes experimentation, testing, and rapid iteration.

Experience

• A minimum of 1 years of relevant experience in machine learning, voice technology, or a related field is preferred.
• Proven track record of solving complex problems with innovative approaches and original thinking.