AI Engineer (Agentic AI & Voice Tech)

AI Engineer (Agentic AI & Voice Tech) Client: Fast-Growing Supply Chain & Logistics Analytics Platform (Confidential) Location: Gurgaon / Gurugram (Work from Office) Experience: 2+ Years Compensation: Up to ₹18 LPA About the Client & Role We are hiring for a fast-growing B2B Supply Chain Analytics & Intelligence platform based in Gurgaon that helps enterprise logistics, procurement, and operations teams automate complex decision-making using real-time data and AI. We are looking for a hands-on AI Engineer with strong production experience in Agentic AI workflows and Speech/Voice technologies (TTS/STT) . You will architect and deploy autonomous AI agents capable of querying supply chain databases, reasoning over operational bottlenecks, and interacting seamlessly with field and enterprise users via low-latency voice and conversational interfaces. Key Responsibilities • Agentic Workflow Engineering: Design, build, and orchestrate multi-agent and tool-calling systems (using frameworks like LangGraph, CrewAI, AutoGen, or custom LLM orchestrators) to automate supply chain workflows, exception handling, and analytics queries. • Voice AI Integration (STT & TTS): Build and optimize real-time conversational voice pipelines utilizing Speech-to-Text (STT) and Text-to-Speech (TTS) models/APIs (e.g., Whisper, Deepgram, ElevenLabs, Sarvam, Cartesia, or WebSockets/WebRTC streaming) for multilingual/Indian-accented logistics and operations use cases. • RAG & Data Grounding: Architect hybrid search and Retrieval-Augmented Generation (RAG) pipelines over structured SQL/ERP data and unstructured supply chain documents to ensure zero-hallucination, deterministic outputs. • Tool Calling & API Integration: Connect LLM agents with internal analytics engines, ERP/WMS/TMS systems, and external APIs using strict schema validation (Pydantic / structured outputs). • Latency, Cost & Eval Optimization: Monitor, benchmark, and optimize end-to-end token latency, voice turn-taking speed, and LLM routing fallbacks in production. Must-Have Requirements (Non-Negotiable) • Experience: 2+ years of hands-on software/AI engineering experience, with proven production deployments in Generative AI and LLM applications. • Agentic AI Expertise: Deep, practical experience building Agentic AI systems —including multi-step reasoning, autonomous tool/function calling, memory management, state machines, and guardrails (LangGraph, LlamaIndex Workflows, CrewAI, or custom Python agent loops). • TTS & STT Knowledge: Hands-on experience working with Text-to-Speech (TTS) and/or Speech-to-Text (STT) models, voice activity detection (VAD), audio streaming pipelines, or real-time voice agents. • Strong Backend & Python Stack: Proficiency in Python (FastAPI / AsyncIO), vector databases (Qdrant, Pinecone, Milvus, or pgvector), and LLM APIs (OpenAI, Anthropic, Gemini, or open-weight models via vLLM). • Structured Data & Analytics Comfort: Ability to build agents that can translate natural language into SQL/analytical queries (Text-to-SQL) or interact with complex B2B data pipelines. Good-to-Have (Preferred) • Prior exposure to Supply Chain, Logistics, E-commerce Ops, FinTech, or B2B Enterprise SaaS analytics. • Experience fine-tuning or optimizing speech models for Indian languages/accents or noisy field environments. • Familiarity with observability and evaluation tools (LangSmith, Langfuse, Arize Phoenix). Why Apply? • High-Impact Engineering: Own end-to-end production AI agents and voice pipelines solving real-world supply chain problems at scale—not just wrapper prototypes. • Compensation: Up to ₹18 LPA , commensurate with hands-on depth in Agentic AI and Voice systems. • Fast-Track Hiring Process: Direct technical discussions with the engineering leadership team.