Lead Software Engineer

Equifax; Thiruvananthapuram; On-site; INR 3,500,000 - 7,000,000; Full time

Job summary

Equifax India is seeking a Lead Software Engineer to serve as the technical anchor and hands-on architect for the enterprise Agentic AI platform. You will design, build, and deploy autonomous, multi-agent architectures capable of reasoning, multi-step planning, tool selection, and robust execution.

In this role, you will lead an engineering squad, establish core engineering standards for AI application development, and write critical-path production code to ensure our AI agents perform safely,

Qualifications

• Architect multi-agent workflows with stateful coordination and planning.
• Develop production-grade AI systems with deterministic behavior.
• Implement secure LLM function-calling interfaces across enterprise systems.
• Lead ADRs, code reviews, and sprint planning with the team.

Responsibilities

• Architect and build stateful multi-agent workflows (ReAct, Plan-and-Solve).
• Create secure interfaces for LLM function calling across ERPs/CRMs.
• Write high-performance orchestration code and optimize context handling.
• Ensure deterministic guardrails and safe AI outputs in production.
• Establish automated evaluation pipelines and deep tracing observability.

Skills

Python (FastAPI)

LLM Applications

Multi-agent orchestration

Docker/Kubernetes

APIs and integration

Performance optimization

Tools

LangGraph

AutoGen

CrewAI

Job description

Equifax India is seeking a Lead Software Engineer to serve as the technical anchor and hands-on architect for our enterprise Agentic AI platform. Moving beyond standard chatbots and retrieval systems, you will design, build, and deploy autonomous, multi-agent architectures capable of reasoning, multi-step planning, tool selection, and robust execution.

In this role, you will lead an engineering squad, establish core engineering standards for AI application development, and write critical-path production code to ensure our AI agents perform safely, deterministically, and efficiently at scale in complex enterprise environments.

What You’ll Do

• Architect and build stateful multi-agent workflows (using ReAct, Plan-and-Solve patterns) where specialized LLM agents delegate tasks, manage short/long-term memory, and synthesize results.
• Build robust, idempotent, and secure interfaces for LLM function-calling across enterprise software databases, ERPs, and CRMs.
• Write clean, highly performant production code for the core orchestration layer, context-window optimization (routing, chunking, compression), and latency/cost management.
• Implement deterministic semantic guardrails around LLM outputs to prevent hallucinations, prompt injection, and unauthorized API/tool calls.
• Establish automated evaluation pipelines (LLM-as-a-judge, deterministic testing) and deep tracing observability to monitor multi-step agent reasoning and prevent infinite loops in production.
• Guide an engineering squad through Architectural Decision Records (ADRs), pair programming, and code reviews while partnering with Delivery Managers on sprint planning.

What Experience You Need

• 8+ years of core software engineering experience, with at least 2+ years building and deploying production-grade LLM applications / Generative AI systems.
• Hands-on experience with multi-agent orchestration frameworks (e.g., LangGraph, AutoGen, CrewAI, or custom graph-based agent state machines).
• Expert proficiency in Python (FastAPI/Pydantic) and experience with vector databases
• Deep understanding of non-deterministic system testing, state management, semantic guardrails (e.g., Guardrails AI, NeMo Guardrails), and prompt security.
• Hands-on experience with containerization (Docker, Kubernetes) and deploying cloud-native AI solutions on AWS, Azure, or GCP.

What Could Set You Apart

• Prior experience building multi-tenant AI platforms in regulated industries (FinTech, Financial Services, Data Analytics).
• Contributions to open-source LLM/Agentic frameworks or publications in applied AI/ML.
• Hands-on knowledge of enterprise fine-tuning, PEFT/LoRA, or custom embedding model training.