About the role

At Networth Corp, we build the AI Gateway and LLMOps layer that lets enterprises deploy GenAI and agents at scale — one gateway, one policy, one bill, real observability. You'll work closely with our AI Red Team & Assurance leads so runtime enforcement matches the policies clients actually need.
The pain is real: every team ships their own LLM prototype, no shared gateway, no cost visibility, no guardrails, security teams saying no by default, and the "second brain" of the enterprise leaks context every third call. Your work makes GenAI production a routine event, not a launch.

What you'll do
• Stand up AI Gateway layers (Databricks AI Gateway, Portkey, LiteLLM, Kong AI Gateway, Azure API Management with AI policies) with unified auth, rate limiting, and cost attribution per team
• Deploy and operate LLM serving infrastructure — vLLM, TGI, Ray Serve, Databricks Model Serving, Azure ML endpoints, Bedrock — with canary rollouts, autoscaling, and traffic mirroring
• Configure guardrails at the gateway: PII redaction, prompt injection defense, jailbreak detection,



output validation, policy enforcement (Guardrails.ai, NVIDIA NeMo Guardrails, Lakera, Protect AI, OWASP LLM Top 10)
• Enable agent platforms with governed tool-use, memory, sandboxed execution, identity propagation (LangGraph, CrewAI, Bedrock Agents, Azure AI Agent Service, Microsoft Copilot Studio, Databricks Mosaic Agent Framework)
• Wire evals into CI so model swaps and prompt changes are gated by regression harnesses (LangSmith, Braintrust, Ragas, DeepEval, Promptfoo, custom golden sets)
• Instrument LLM traffic: latency, token cost, cache hits, quality drift, safety flags, jailbreak attempts (Langfuse, Arize Phoenix, WhyLabs, LangSmith, OpenTelemetry GenAI semantic conventions)
• Own RAG operations — chunking pipelines, embedding refresh, vector DB deployment, hybrid search tuning
• Build model gateway routing for cost / latency / quality tradeoffs — small model first, escalate to Claude

📌 LLMOps Engineer (India)
🏢 Networth
📍 India