Data Scientist NLP & Conversational AI

You are looking for a Data Scientist specializing in NLP and conversational AI to join a core team that focuses on real-time voice intelligence. In this role, you will be responsible for building systems that involve intent detection, generative reasoning, and speech/audio insights to develop responsive and persuasive AI callbots. This position is suited for individuals who excel in applied NLP and voice technology and are passionate about creating scalable AI systems that are natural and human-like. As a Data Scientist specializing in NLP & Conversational AI, you will be based in Gurugram and should have at least 2 years of experience. This is a full-time position where you will work collaboratively with the engineering teams to deploy NLP models in real-time inference systems. Your responsibilities will include building and refining models for intent recognition, speech act classification, and response routing. Additionally, you will design lightweight NLP classifiers and heuristics to optimize LLM usage, and enhance conversational dynamics using various techniques such as turn-taking prediction and pause detection. To excel in this role, you should have proficiency in NLP frameworks such as HuggingFace Transformers, spaCy, NLTK, fastText, and Sentence Transformers. Familiarity with ML libraries like scikit-learn, PyTorch, TensorFlow, and LightGBM is essential. Experience in audio signal processing techniques like MFCCs, VAD, filler detection, and silence segmentation would be beneficial. Knowledge of real-time or low-latency model design, RAG pipelines, dialog state management, dialogue act tagging, and conversational UX design is also required. Preferred or bonus skills that would be advantageous include familiarity with Whisper, Coqui, Bark, or other open-source STT/TTS models, prompt engineering, LLM optimization, and experience with streaming inference architectures or edge AI. Exposure to hybrid response generation systems and the use of experiment tracking tools like MLflow or Weights & Biases would be a plus. The ideal candidate should be comfortable working in fast-paced and ambiguous environments, possess startup or early-stage product experience, and showcase a strong applied portfolio with GitHub repositories, notebooks, demos, or a track record in Kaggle/NLP competitions. Additionally, having an eagerness to develop production-ready, real-time ML features along with qualities like curiosity, creativity, and a collaborative mindset are highly valued for this role.,