Fellowship : (Agent Intelligence & Evaluations)

Design and develop advanced evaluation frameworks tailored for voice agents, including audio-native metrics and adversarial conversational datasets.

Build end-to-end observability tools correlating audio, speech-to-text, LLM reasoning, and tool outputs for root cause analysis across interaction cascades.

Develop self-improvement systems by mining failure patterns from production traces and implementing targeted fine-tuning and validation under adversarial conditions.

Proficiency in Python and familiarity with either speech models (e.g., Whisper, Conformer), LLM tool-use/agent frameworks, or observability stacks (e.g., OpenTelemetry).

Relevant research experience demonstrated by publications at conferences such as Interspeech, ACL, NeurIPS, or EMNLP in speech, dialogue systems, or human-AI interaction.

Currently enrolled PhD student in ML, NLP, or speech, or exceptional MS students/research engineers with publication track record.

Work Experience Required: Not explicitly mentioned in the JD.

Research-oriented with proven ability to publish peer-reviewed work at top conferences in speech or NLP domains.

Experience bridging research and production by shipping impactful tooling or frameworks relevant to voice agent evaluation or observability.

Background or interest in real-time systems, telephony, streaming pipelines, or advanced evaluation methodologies and interpretability.