AI Scientist

Role Overview: You will be responsible for the end-to-end development of production-grade ASR and Speech AI systems for real-time voice agents at Uniphore. Your focus will be on designing and optimizing streaming architectures, including end-pointing, end-of-turn detection, and conversational turn-taking. Additionally, you will work on LLM fine-tuning, building Retrieval-Augmented Generation (RAG) pipelines, and developing agentic systems for enterprise conversational AI use cases. Collaboration with engineering teams to deploy scalable, low-latency, and cost-efficient speech/NLP systems will also be a key part of your role. Key Responsibilities: - End-to-end development of production-grade ASR and Speech AI systems for real-time voice agents. - Design and optimize streaming architectures including end-pointing, end-of-turn detection, and conversational turn-taking. - Focus on LLM fine-tuning and building Retrieval-Augmented Generation (RAG) pipelines as well as agentic systems for enterprise conversational AI use cases. - Drive integration/benchmarking of allied modules such as VAD, LID, speaker diarization, and paralinguistic modeling (emotion, prosody). - Own benchmarking strategy against 3P ASR/LLM and define evaluation frameworks (accuracy, latency, cost). - Collaborate with engineering teams to deploy scalable, low-latency, and cost-efficient speech/NLP systems. Qualifications: - Master's or Ph.D in Computer Science, Electrical/Computer Engineering or related field. - Minimum of around 2 years of industry experience in AI, NLP, Vision or ASR for candidates with Master's degree. - Expertise in Python and deep learning frameworks such as PyTorch or TensorFlow, with experience using modern ML tooling (e.g., Hugging Face, vLLM). - Experience optimizing streaming ASR (RTF, decoding strategies, end-pointing latency). - Ability to drive benchmarking, system optimization, and deployment readiness. - Good publication record in leading conferences. - [Good to have] Prior Experience in pre-training, fine-tuning, agentic systems and reinforcement learning. Role Overview: You will be responsible for the end-to-end development of production-grade ASR and Speech AI systems for real-time voice agents at Uniphore. Your focus will be on designing and optimizing streaming architectures, including end-pointing, end-of-turn detection, and conversational turn-taking. Additionally, you will work on LLM fine-tuning, building Retrieval-Augmented Generation (RAG) pipelines, and developing agentic systems for enterprise conversational AI use cases. Collaboration with engineering teams to deploy scalable, low-latency, and cost-efficient speech/NLP systems will also be a key part of your role. Key Responsibilities: - End-to-end development of production-grade ASR and Speech AI systems for real-time voice agents. - Design and optimize streaming architectures including end-pointing, end-of-turn detection, and conversational turn-taking. - Focus on LLM fine-tuning and building Retrieval-Augmented Generation (RAG) pipelines as well as agentic systems for enterprise conversational AI use cases. - Drive integration/benchmarking of allied modules such as VAD, LID, speaker diarization, and paralinguistic modeling (emotion, prosody). - Own benchmarking strategy against 3P ASR/LLM and define evaluation frameworks (accuracy, latency, cost). - Collaborate with engineering teams to deploy scalable, low-latency, and cost-efficient speech/NLP systems. Qualifications: - Master's or Ph.D in Computer Science, Electrical/Computer Engineering or related field. - Minimum of around 2 years of industry experience in AI, NLP, Vision or ASR for candidates with Master's degree. - Expertise in Python and deep learning frameworks such as PyTorch or TensorFlow, with experience using modern ML tooling (e.g., Hugging Face, vLLM). - Experience optimizing streaming ASR (RTF, decoding strategies, end-pointing latency). - Ability to drive benchmarking, system optimization, and deployment readiness. - Good publication record in leading conferences. - [Good to have] Prior Experience in pre-training, fine-tuning, agentic systems and reinforcement learning.