Role: Agentic AI Developer

Experience:
5–8 years in Python-first software development; hands-on AI/agentic development experience required Notice period: Only Immediate
- 15days

Location:
Anywhere in India Shift Timings: 10 am to 8 PM A hands-on Python developer who builds, tests, and maintains the agents, pipelines, and AI-enabled features of the IPS insurance broking platform. This is a coding role — candidates are expected to write production-quality Python every day, working within the Microsoft Agent Framework multi-agent pipeline running on Azure. 4D band targets developers who can own a feature end-to-end with light oversight; 4C band targets developers who can deliver well-defined tasks within the platform with close collaboration. Responsibilities Agentic AI Development (Primary Focus) Build and maintain agents within the Microsoft Agent Framework multi-agent pipeline — including workflow design in the visual designer, prompt engineering, tool/function registration, and agent-to-agent handoff logic. Write Python services (FastAPI) that form the backend of the agentic platform — including async patterns, Azure SDK for Python, and RESTful API design. Integrate Azure AI services into the platform: Azure OpenAI (LLM invocation, token budgeting, retry/fallback), Azure AI Foundry (model deployment, versioning), Azure AI Content Understanding (document extraction schemas, classifier configuration, confidence scoring), Azure AI Translator, and Azure AI Search (RAG, vector-search patterns). Implement rules engine logic for hard-check and soft-check validation rules covering insurance broking workflows — policy search, co-broker duplicate detection, mid-term binding, carrier-portal submission. Develop Human-in-the-Loop workflow components: exception queues, reviewer edit flows, escalation paths. Platform Engineering Write clean, version-controlled, well-tested Python code — unit tests, integration tests, and tolerance-based assertions for non-deterministic LLM outputs. Integrate with insurance broking systems via REST APIs: AIM API, ImageRight, carrier portals, and Cora Sequence — including authentication patterns (OAuth2, managed identities).



Contribute to CI/CD pipelines for the platform (Azure DevOps or GitHub Actions); follow DevSecOps practices. Participate in code reviews, sprint planning, and retrospectives; maintain technical documentation. Debug and resolve production issues in the agentic pipeline; contribute to root-cause analysis. Minimum Qualifications Solid, hands-on Python (PRIMARY language — non-negotiable): FastAPI, async/await, Pydantic, Azure SDK for Python, RESTful API development, clean code practices. Must be able to write production Python from scratch in the interview, not just describe it. Microsoft Agent Framework — hands-on development experience building agents, registering tools/functions, and designing multi-agent workflows. Candidates with Semantic Kernel experience are eligible; migration to Microsoft Agent Framework is in progress on the platform. Azure OpenAI — LLM API invocation, prompt construction, token budgeting, retry/fallback strategies, structured output parsing. Azure platform fundamentals: Azure Key Vault (secrets), managed identities / Microsoft Entra ID (RBAC), Azure Storage, Azure Monitor / Application Insights — sufficient to develop and deploy features on the platform independently. REST API integration: consuming external REST APIs with proper authentication (OAuth2, API keys, managed identities); JSON handling; OpenAPI/Swagger contract familiarity. Git — version control, branching, code review workflow; CI/CD basics (Azure DevOps or GitHub Actions). Unit and integration testing — pytest or equivalent; writing tests for deterministic and non-deterministic (LLM) outputs. 4D Band — Additional Expectations 5–8 years of experience; able to own a feature end-to-end from design through production deployment with light oversight from the Tech Architect. Azure AI Foundry — model deployment management, versioning, and guardrails configuration. Strong understanding of multi-agent pipeline design: agent orchestration,



tool registration, context window management, and prompt chaining strategies. Able to contribute to low-level design documents and integration pattern decisions; leads code review for 4C developers. Preferred Qualifications / Skills Azure AI Content Understanding — hands-on experience building document extraction schemas, configuring classifiers, and interpreting confidence scores; strongly preferred for this role given the CRC/IPS document processing pipeline. Azure AI Search — RAG pipeline implementation, vector-search index design, hybrid retrieval strategies. Azure AI Translator — document translation integration within an agentic pipeline. Insurance broking domain awareness: policy search, AIM, carrier portals, co-broker duplicates, mid-term binding — domain exposure is a significant plus; not a hard requirement for developers (the BA role carries domain ownership). Java familiarity (Spring Boot) — sufficient to read and debug the Java platform layer when integration issues span services; not required for the AI/agentic coding work. Event-driven patterns: Azure Service Bus, Kafka, or RabbitMQ for async messaging between platform components. MLOps basics: model versioning, deployment pipelines, monitoring of AI model outputs in production. Docker; Kubernetes basics — sufficient to containerise a Python service and understand how it runs in the platform. SECONDARY — Strong Plus: Azure AI Content Understanding — document extraction schemas, classifier config, confidence scoring (strongly preferred) Azure AI Foundry — model deployment, versioning, guardrails (required at 4D band) Azure AI Search — RAG pipeline, vector-search index design, hybrid retrieval Azure AI Translator — document translation in agentic workflows Insurance broking integrations — AIM API, ImageRight, carrier portals, Cora Sequence REST Azure Service Bus / Kafka — async messaging between platform components OPTIONAL — Nice-to-Have: Java / Spring Boot familiarity — to read and debug the platform layer Docker; Kubernetes basics MLOps — model versioning, deployment pipelines, production AI monitoring AWS or GCP familiarity (Azure is the required platform)

📌 Agentic AI Developer (Bengaluru)
🏢 SysTechCorp
📍 Bengaluru