Senior AI/ML Engineer - Agent Development

Hiring : Senior AI/ML Engineer

Experience : 6+ Years

Location : Bangalore & Noida

Mode of Work : Hybrid

Summary :

We are looking for a Senior AI/ML Engineer to join an Agent Development Squad building AI-powered financial close automation. You will own the LLM and agent layer, designing and building agent workflows using LangChain/LangGraph, integrating with Azure OpenAI or other frontier LLMs, and ensuring every agent action is traceable, evaluable, and production-grade. This is a hands-on engineering role, not a research role. The platform is Python-first, and you will work alongside backend and frontend engineers in a cross-functional squad.

What You'll Do :

- Design and build LLM-based agent workflows using LangChain and LangGraph, covering multi-step graphs, tool use, agent handoffs, and state management.

- Integrate and optimize Azure OpenAI or other frontier LLM providers (Anthropic Claude, Google Gemini, etc.), managing prompt design, token usage, latency, and model versioning.

- Build and maintain RAG pipelines covering document ingestion, embedding generation, vector search using pgvector or equivalent vector database, and retrieval-augmented generation for financial document reasoning.

- Own LLM observability using Langfuse, instrument agent traces, build evaluation harnesses, track confidence scores, and detect model regressions.

- Collaborate with the Software Engineering Architect on agent orchestration patterns and inter-agent communication contracts.

- Contribute to confidence threshold calibration, defining the auto-resolve vs. human-escalation boundary per agent.

- Ensure all agent actions produce structured, auditable outputs compatible with the platform's SOX-compliant governance framework.

Who You Are :

AI / LLM Engineering - Primary Gate :

- 6+ years of total software engineering experience with 3+ years building and shipping production-grade LLM-based applications.

- Proven experience building production-grade agents, both assistive (human-in-the-loop, approval-driven) and ambient (autonomous, background execution), using LangChain, LangGraph, and LangServe; knows when to use each and how they compose.

- Familiarity with agent communication protocols, specifically Model Context Protocol (MCP) and Agent-to-Agent (A2A) and working knowledge of other Agent Development Kits such as Google ADK, AutoGen, CrewAI, or Semantic Kernel.

- Experience integrating with Azure OpenAI or other frontier LLM provider APIs (Anthropic Claude, Google Gemini, Mistral, etc.), including structured outputs, function calling, token management, and latency optimization.

- RAG pipeline development with hands-on experience in at least one vector database (pgvector, Pinecone, Weaviate, Qdrant, FAISS, or equivalent), covering embedding pipelines and retrieval strategies.

- Prompt lifecycle management and eval pipelines, covering versioning, rollback, environment-specific configuration, and running evaluation harnesses to regression-test prompt changes and validate RAG output.

- Experience building reliable production agent systems with input/output guardrails, confidence thresholding, fallback handling, and full traceability of every decision and tool call via Langfuse or equivalent.

Backend Engineering :

- Python is your primary language; proficient with FastAPI or equivalent async frameworks for building agent service APIs.

- PostgreSQL with pgvector or equivalent vector database (Pinecone, Weaviate, Qdrant, FAISS), including schema design for agent state, audit logs, and vector search.

- Docker and Kubernetes for containerized service deployment in production environments.

- Azure DevOps CI/CD for integrating AI pipelines into automated build and deployment workflows.

Nice to Have :

- Experience with financial domain data such as accounting entries, reconciliations, accruals, or variance analysis.

- Familiarity with SOX/GDPR compliance requirements for AI decision audit trails.

- Contributions to open-source LLM or agent framework projects.

- Experience with federated learning or model fine-tuning pipelines.

What You'll Learn & Gain :

- End-to-end ownership of the LLM and agent layer on a live enterprise financial platform used by global customers.

- Deep hands-on experience with LangChain/LangGraph, Azure OpenAI and other frontier LLMs, agent protocols (MCP, A2A), and Langfuse in a production context where quality, auditability, and latency all matter.

- Exposure to building AI in a high-compliance, SOX-governed environment, balancing model performance with explainability and audit requirements.

- Collaboration with backend, frontend, and platform engineers in a squad model where your technical decisions directly shape product behavior.