Senior AI Engineer - Voice AI & Agentic Systems

Danish Mullaji; Gurugram District; On-site; INR 4,500,000 - 7,500,000; Full time

Job summary

Danish Mullaji in Gurgaon is seeking a hands-on Senior AI Engineer to own the intelligent core of our agent-assist stack. You will develop real-time Voice AI systems, LangChain-based workflows, and production ML pipelines for multilingual conversations at scale.

You will ship agent-assisted features, optimize latency and throughput, and ensure robust deployments using Docker and Kubernetes in a fast-paced startup setting.

Qualifications

• Production Voice AI experience with real-time conversational systems.
• Hands-on with LangChain/LangGraph for agentic AI orchestration.
• Production-grade Python development with REST APIs (FastAPI).
• Deployment of ML/AI systems at scale using Docker and Kubernetes.

Responsibilities

• Architect and ship real-time voice AI and agentic workflows in production.
• Build scalable ML pipelines with prompt engineering and RAG architectures.
• Fine-tune models for Indian accents and vernacular languages.
• Deploy AI services on Kubernetes with low latency and high concurrency.
• Integrate AI modules with Dialer, CRM, IVR, and agent tools.

Skills

Production Voice AI

Agentic AI & LLM

Python & FastAPI

Docker & Kubernetes

Tools

LangChain

LangGraph

FastAPI

Docker

Kubernetes

Job description

Senior AI Engineer Voice Bot & Agentic AI

Gurgaon | Full-Time | 6-9 Years Experience ( 3+ years in production AI/ML)

We're building the next-generation AI-native, multilingual, omnichannel communication platform for emerging markets.

We're looking for a hands-on Senior AI Engineer who will own and deliver the intelligent core of our agent-assist stack including real-time Voice AI, agentic workflows, LLM-based automation, and production ML pipelines that power conversations at scale.

What You'll Build

This is a foundational role where you'll architect and ship production AI systems that directly impact how businesses communicate with customers across India and beyond.

You'll Own

• Real-time Voice AI Systems — STT LLM reasoning TTS loops with sub-second latency
• Agentic Workflows — Multi-step decision engines using LangChain/LangGraph for call summarization, sentiment detection, auto-disposition, and escalation routing
• Production LLM Pipelines — Prompt engineering, RAG architectures, context management, and evaluation frameworks
• Model Optimization — Fine-tune speech and language models for Indian accents and vernacular languages, including Hindi, Tamil, Telugu, and Bengali
• Scalable Inference — Deploy AI services on Kubernetes with FastAPI, optimized for high-concurrency, low-latency production environments
• Platform Integration — Connect AI modules into Dialer, CRM, IVR, and agent-assist tools

You'll Ship

• Conversational AI that handles real customer interactions, not just demos
• Agent-assist features that save agents 30+ seconds per call
• Automated workflows that route, tag, and resolve issues without human intervention
• ML systems that learn from multilingual, noisy, real-world voice data

Must-Have Skills

We'll assess these deeply.

Mandatory

1. Production Voice AI Experience

Real-time conversational systems (STT LLM TTS), not just transcription or batch audio.

You should have built voice bots or agents that interact with real users in production.

2. Agentic AI & LLM Orchestration

Hands-on experience with LangChain, LangGraph, or similar frameworks, building agents that:

• Reason over context
• Use tools
• Maintain state
• Make autonomous decisions

We're looking for genuine agentic AI experience — not just chaining API calls.

3. Python & FastAPI Production Development

You write production-grade Python and have built and deployed REST APIs using FastAPI.

You should understand:

• Async programming
• Error handling
• Logging and observability
• API design
• Production debugging

4. ML/AI Deployment at Scale

You've shipped models to production using Docker + Kubernetes and understand CI/CD for ML systems.

You should have experience with:

• Debugging latency
• Optimizing throughput
• Monitoring inference in production
• Scaling AI workloads

Strong Preference

Not dealbreakers, but highly valued:

• Speech AI — Whisper, Deepgram, or equivalent STT/TTS systems in production
• Contact Center / IVR / Dialer / CRM domain knowledge
• Indian language models — Hindi, Tamil, Telugu, Bengali, or Hinglish ASR/NLP
• Data infrastructure — PostgreSQL, Redis, Kafka for real-time pipelines
• HuggingFace ecosystem and model fine-tuning — LoRA/QLoRA, quantization

Nice to Have

• Experience with Rasa, Coqui TTS, or open-source Voice AI stacks
• Prior experience in SaaS startups, especially communication/telephony
• Speech emotion recognition, speaker diarization, or acoustic modeling
• Compliance-aware AI — DPDP, GDPR-ready data handling

You'll Thrive Here If You

• Have 6–9 years of software/AI engineering experience, with at least 3+ years in production AI/ML across LLMs, Voice AI, NLP, or Conversational Systems
• Have shipped AI systems that real users depend on — not just POCs or Jupyter notebooks
• Are highly hands-on — you write code, review PRs, debug production issues, and own deployments
• Understand production trade-offs — latency vs. accuracy, cost vs. quality, speed vs. polish
• Can work effectively in a startup environment — ambiguity is opportunity, scrappiness is valued, and ownership is expected
• Communicate clearly and can explain AI systems to product, engineering, and business teams

You don't need to know everything on Day 1 — but you should be comfortable learning fast.

Why This Role Matters

You'll be shaping how millions of users interact with businesses — in their language, on their terms, with AI that actually understands context and intent.

This isn't a research lab. This is production AI at the edge of what's possible today.

You'll work on:

• Multilingual Voice AI for Tier 2/3 India, where English-first solutions often fail
• Affordable, fast-deploy SaaS for SMBs that can't afford enterprise CCaaS
• Real-time agent assistance that makes human agents up to 3 more effective

You'll work with modern AI tools such as GPT, Whisper, and LangChain while solving hard, real-world problems including:

• Noisy audio
• Code-mixed languages
• Low-latency constraints
• Cost optimization
• Real-time inference at scale