Senior Ai Engineer Voice Ai Agentic Systems Danish Mullaji Gurugram

Vibehackers; Gurugram District; On-site; INR 4,000,000 - 7,000,000; Full time

Job summary

Vibehackers in Gurugram seeks a Senior AI Engineer specializing in Voice and Agentic AI to design and ship production-grade systems for multilingual, real-time conversations. You will own STT/LLM/TTS pipelines, agent workflows, and scalable inference, deploying on Kubernetes with FastAPI and integrating with Dialer, CRM, IVR.

The role requires 6-9 years in software/AI with 3+ years production experience, plus hands-on LangChain/LangGraph, Python, Docker/Kubernetes, and Indian language AI

Qualifications

• Must have shipped production AI systems used by real users.
• Hands-on experience with real-time voice AI systems (STT/LLM/TTS).
• Experience with LangChain/LangGraph or similar orchestration for autonomous decisions.

Responsibilities

• Design and ship real-time voice AI systems with sub-second latency.
• Build agentic workflows and multi-step decision engines.
• Develop production LLM pipelines with prompt engineering and context management.
• Fine-tune models for Indian accents and vernacular languages.

Skills

Python

FastAPI

LangChain

Docker

Kubernetes

Tools

PostgreSQL

Redis

Kafka

Job description

Senior Ai Engineer Voice Ai Agentic Systems Danish Mullaji Gurugram

Uses LangChain, GPT, and Whisper to build agentic voice AI and real-time inference; focused on production-grade AI systems rather than prototypes.

About the Role

Senior AI Engineer to design and ship production-grade voice and agentic AI systems for a multilingual omnichannel communication platform, focusing on real-time STT/LLM/TTS pipelines, agent workflows, and scalable inference. The role owns deployment, optimization, and integration of AI services that power customer conversations at scale in India.

Job Description

Role

Senior AI Engineer (Voice & Agentic AI) responsible for architecting and delivering the intelligent core of an agent-assist stack: real-time Voice AI (STT, LLM reasoning, TTS), agentic multi-step workflows, LLM-based automation, and production ML pipelines for multilingual, noisy voice data.

Key Responsibilities

• Design and ship real-time voice AI systems with sub-second latency (STT -> LLM reasoning -> TTS loops).
• Build agentic workflows and multi-step decision engines using LangChain/LangGraph (call summarization, sentiment detection, auto-disposition, escalation routing).
• Develop production LLM pipelines: prompt engineering, RAG architectures, context management, and evaluation frameworks.
• Fine-tune speech and language models for Indian accents and vernacular languages (Hindi, Tamil, Telugu, Bengali, Hinglish).
• Deploy and scale inference services on Kubernetes with FastAPI; optimize for high concurrency and low latency.
• Integrate AI modules with Dialer, CRM, IVR, and agent-assist tools.
• Monitor, debug, and optimize production ML systems for latency, throughput, and cost.

Requirements

Mandatory

• 6-9 years of software/AI engineering experience with at least 3+ years in production AI/ML.
• Proven production experience building real-time conversational voice systems (STT, LLM, TTS) interacting with real users.
• Hands-on experience with agentic AI/LLM orchestration (LangChain, LangGraph, or similar) that reasons, uses tools, maintains state, and makes autonomous decisions.
• Strong Python development skills and experience building/deploying REST APIs with FastAPI; familiarity with async programming, error handling, logging/observability, and production debugging.
• Experience deploying models and services to production using Docker and Kubernetes and familiarity with CI/CD for ML systems.

Strong Preference

• Experience with Whisper, Deepgram, or equivalent STT/TTS systems in production.
• Contact center / IVR / Dialer / CRM domain knowledge.
• Experience with Indian language models and ASR/NLP for Hindi, Tamil, Telugu, Bengali, or Hinglish.
• Familiarity with PostgreSQL, Redis, Kafka for real-time pipelines.
• Experience with Hugging Face ecosystem, fine-tuning (LoRA/QLoRA), and quantization.

Nice to Have

• Experience with Rasa, Coqui TTS, or other open-source voice AI stacks.
• Prior SaaS/telephony startup experience.
• Experience in speech emotion recognition, speaker diarization, or acoustic modeling.
• Knowledge of compliance-aware data handling (DPDP, GDPR-ready).

Experience & Expectations

• Must have shipped production AI systems used by real users (not only prototypes or research).
• Comfortable owning code, PR reviews, production debugging, and deployments in a startup environment.
• Ability to balance latency vs. accuracy and cost vs. quality trade-offs.

Why this role matters

You will shape multilingual, real-world conversational AI for emerging markets, enabling more effective human-agent interactions and scalable automated workflows. The work emphasizes production readiness, low-latency inference, and robust multilingual performance.

Skills

Production Deployment System Design Async Programming API Design Error Handling Logging and Observability Production Debugging CI/CD for ML Performance Optimization Prompt Engineering Model Fine-tuning Multilingual NLP Agentic AI Orchestration Cross-functional Communication Ownership