Role description:

Employment Type: Full Time

Job Location: Mumbai

Work Mode: Hybrid

Role Category: AI Developer

Required Skill Set: Experience fine-tuning open-source LLMs or SLMs (Small Language Models) specifically for high-speed conversational execution. Prior experience building omnichannel systems where a single conversational engine powers both a text-based web/WhatsApp chatbot and a telephonic voice agent smoothly.

Experience: 1-2 Year

Job Summary:

We are looking for a Senior AI Engineer to spearhead the design, architecture, and deployment of our next-generation conversational AI ecosystems. In this role, you will bridge the gap between advanced Large Language Models (LLMs), Agentic workflows, and real-time voice infrastructure. You wont just be plugging into basic APIs; you will be optimizing low-latency audio pipelines, managing complex conversational state machines, and building intelligent agents that can handle unstructured, real-world human interactions seamlessly over both voice and chat interfaces.

Key Responsibilities:

Core Responsibilities

- End-to-End Voice Architecture: Design, build, and optimize scalable pipelines integrating Automatic Speech Recognition (ASR), Text-to-Speech (TTS), and LLMs for ultra-low latency less than 1.5s perceived response time voice bots.

- Agentic Framework Design: Develop and deploy multi-agent orchestration workflows that allow bots to reason, call external APIs, query databases dynamically, and execute complex business logic during live conversations.

- State Context Management: Implement robust dialog management and context retention systems, ensuring the bot can handle interruptions, digressions, and multi-turn conversations without losing track.

- Backend API Integration: Build highly performant backends to connect conversational interfaces with enterprise CRMs, databases, and third-party systems using robust API architectures.

- Evaluation Optimization:



Set up continuous evaluation frameworks to monitor bot performance, natural language understanding (NLU) accuracy, voice latency, and fallback rates, iteratively tuning prompts and fine-tuning models to improve user experience.

Technical Profile Requirements

- Conversational Voice AI Expertise (Primary) Deep understanding of Telephony WebRTC integration (e.g., Twilio, LiveKit, Vapi, Daily.co) for streaming audio.

- ASR (Speech-to-Text) engines (Deepgram, Whisper, AssemblyAI) for real-time streaming, noise filtering, and accents.

- TTS (Text-to-Speech) systems (ElevenLabs, Cartesia, OpenAI Voice) focusing on emotional inflection, prosody, and low-latency audio synthesis.

- barge-in management (handling user interruptions mid-speech) and silence detection.

- Generative AI Agentic Workflows

- Expertise in orchestrating LLMs (OpenAI GPT-4o, Claude 3.5 Sonnet, open-source Llama models) for conversational tasks.

- LangChain, LangGraph, or CrewAI to build autonomous, tool-using AI agents.

- RAG (Retrieval-Augmented Generation) techniques and Vector Databases (e.g., Pinecone, pgvector, Milvus) for grounding bot responses in specific datasets.

- Python (for AI framework integration and data processing).

- FastAPI experience for asynchronous APIs, or robust backend architectures using up-to-date web frameworks.

- MySQL, PostgreSQL and NoSQL systems for state persistence and logging.

- WebSockets and gRPC for bidirectional, real-time data streaming.

- Soft Skills Best Practices

- Conversation Design Mindset Ability to collaborate on structuring natural, empathetic, and efficient prompt designs that avoid robotic, cyclical loops.

- Production Discipline Experience with CI/CD, containerization (Docker), logging/observability tools (LangSmith, Arize, or Prometheus), and managing AI infrastructure at scale.

Disclaimer: This job posting has been aggregated from external source. Role details, content, and availability are subject to change. Applicants are advised to confirm the latest information directly on the company website before applying.

📌 AI Developer (Mumbai)
🏢 Byasa Tech Solutions
📍 Mumbai