Back
Niche academic ML fellowship attracts PhD applicants and research engineers, producing moderate competition.
Highly specialized ML/speech research skills and publication history make cross-industry transfer limited.
Requires PhD-level research track record, publications, and specific speech/LLM skills, so filters are stringent.
Login to See Your Match Score
Create a free account or log in to unlock your CV match score across:
Develop audio-native evaluation metrics and adversarial conversational datasets for voice agent performance assessment.
Design observability schemas and analysis layers to trace and diagnose multi-component voice agent failure cascades.
Implement closed-loop self-improvement systems by mining failure patterns and validating targeted fine-tuning or prompt updates.
Experience: Currently enrolled PhD students in ML, NLP, or speech preferred; exceptional MS students or research engineers with publication records also considered.
Technical skills: Proficiency in Python; familiarity with speech models (e.g., Whisper, Conformer), LLM tool-use/agent frameworks, or observability stacks (e.g., OpenTelemetry, Langfuse).
Research background: Publications at venues like Interspeech, ACL, NeurIPS, or EMNLP on relevant topics such as speech, dialogue systems, or human-AI interaction.
Compensation and term: 4-month fellowship with ₹50,000/month stipend.
Research-focused with a track record of publishing in top speech, NLP, or ML conferences, showing deep domain expertise.
Ability to integrate research outputs into deployable systems that improve real-world voice agent performance.
Comfortable working with complex pipelines involving speech recognition, large language models, and multi-modal observability frameworks.