Technical Proficiency :
- Strong programming in Python, TensorFlow / PyTorch, and model APIs (Hugging Face, LangChain, OpenAI, etc.).
- Expertise in STT, TTS, S2S, speaker diarization, and speech emotion recognition.
- LLM fine-tuning, model optimization (quantization, distillation), RAG pipelines.
- Understanding of agentic frameworks, cognitive architectures, or belief-desire-intention (BDI) models.
- Familiarity with Edge AI deployment, low-latency model serving, and privacy-compliant data pipelines.
Desirable :
Exposure to agent-based simulation, reinforcement learning, or behavioralmodeling.Publications, patents, or open-source contributions in conversational AI or GenAI systems.Role & Responsibilities
Agentic AI Development :
Design and develop multi-agent conversational frameworks with adaptive decision-making capabilities.Integrate goal-oriented reasoning and memory components into agents using transformer-based architectures.Build negotiation-capable bots with real-time context adaptation and recursive feedback processing.Generative AI & Model Optimization :
Fine-t une LLMs / SLMs using proprietary and domain-specific datasets (NBFC, Financial Services, etc.).Apply distillation and quantization for efficient deployment on edge devices.Benchmark LLM / SLM performance on server vs. edge environments for real-time use cases.Speech and Conversational Intelligence :
Implement contextual dialogue flows using speech inputs with emotion and intent tracking.Evaluate and deploy advanced Speech-to-Speech (S2S) models for naturalistic voice responses.Work on real-time speaker diarization and multi-turn, multi-party conversation tracking.Voice Biometrics & AI Security :
Train and evaluate voice biometric models for secure identity verification.Implement anti-spoofing layers to detect deepfakes, replay attacks, and signal tampering.Ensure compliance with voice data privacy and ethical AI guidelines.ref : hirist.tech)