AI Application Engineer

Position Overview

Location: On-site/Hybrid (Delhi)

Team: AI & Intelligent Systems

Seniority: Mid-Senior / Lead

Type: Full-time

Who We're Looking For

We're not hiring a backend engineer or a data scientist.

We're hiring an AI-Native Builder — someone who can design, build, and deploy:

• LLM-powered features (Copilots, Autopilots, Internal Assistants)

• Agents that automate operations, workflows, or decision-making

• ML+AI orchestration systems (LangChain, LangGraph, AutoGen, etc.)

• AI-first infra that learns, adapts, and scales with usage

This role is perfect for someone who has:

• Built production systems powered by LLMs or lightweight ML models

• Worked with agent frameworks and has real opinions on prompts, memory, planning, and evaluation

• Shipped AI features into production SaaS/Fintech environments

• Strong backend (Python, FastAPI, Postgres, async, queues) with MLOps comfort

What You’ll Be Doing

• Design and implement agentic workflows (e.g. Payment Reconciliation Agents, Fraud Triage Copilots,

Internal Audit Chatbots)

• Build AI-powered orchestration systems — where LLMs make smart routing/fallback/retry decisions in

payments

• Develop domain-specific copilots for our internal teams: sales, finance, compliance, etc.

• Collaborate with our product and infra teams to deeply integrate AI into core payment operations

• Architect and own internal tools powered by OpenAI, Claude, Ollama, or fine-tuned domain models

Examples of Projects

• AI Payment Orchestrator: Route UPI/netbanking/card payments to optimal processors using success

rates, cost models, and fallback strategies – controlled by LLM-powered logic instead of static code

• Compliance CoPilot: An agent that reads PDF guidelines from RBI and tags risk items in real-time

merchant onboarding

• Intelligent Dashboard Agents: Summarize thousands of transactions with charts and anomalies in a

Slack message

• AutoPilot for Ops: Natural-language-triggered workflows that send emails, raise tickets, block

suspicious accounts, or escalate

Tech Stack

• LLMs: OpenAI (GPT-4/4o), Claude, Ollama, Gemini, fine-tuned transformers

• Frameworks: LangChain, LangGraph, AutoGen, FastAPI, Celery

• Infra: Docker, Kubernetes (EKS), Redis, Postgres, Snowflake

• Dev: GitHub, JIRA, Slack, Confluence, VSCode, Make.com (for AI RPA)

• MLOps (Basic): Weights & Biases, HuggingFace, vector DBs like Qdrant or FAISS

You Should Apply If...

• You've built something real with AI (copilots, bots, assistants, agents)

• You can go from natural language to working backend

• You love orchestrating AI agents, not just calling.chat_completion()

• You have strong reasoning ability, product sensibility, and know how to move from POC → Production

• You can ship fast, iterate, and thrive in chaos

Bonus Points

• Prior experience in payments, banking, or regulatory tech

• You’ve built open-source LLM tools or agents

• You’ve experimented with multi-agent frameworks (AutoGen, CrewAI, etc.)

• You’ve used Claude or GPT-4 to generate AND deploy working code

• You’ve orchestrated live LLM-in-the-loop agents for real-world workflows

Thanks

Team-HR