Forward Deployed Engineer - Data / Reporting-Alabang

iQor; Muntinlupa; Hybrid; PHP 1,200,000 - 1,800,000; Full time

Job summary

iQor is seeking an experienced AI/ML engineer in Metro Manila to design, build, and deploy enterprise-grade AI agents using code-first frameworks. You will focus on architecture, tool integration, grounding, and secure identity patterns to deliver production-ready agents with traceable grounding and human-in-the-loop controls.

You will work with Semantic Kernel, LangGraph, LangChain, and Autogen, building end-to-end RAG pipelines, integrating with SharePoint/Dataverse, and deploying via Teams

Qualifications

• 5+ years of professional AI/agent development experience.
• Proven delivery of at least two production AI/agent apps.
• Strong Python and/or C#/.NET with clean API design.
• Hands-on with an agent framework (Semantic Kernel, LangGraph, LangChain, Autogen).
• Deep prompt engineering: system prompts, few-shot patterns and structured outputs.
• Experience with RAG end-to-end and vector stores.
• Experience in tool/function calling design and error handling.
• Enterprise identity and API integration know-how (Entra ID, OAuth 2.0).

Responsibilities

• Design agent architectures for single-agent, multi-agent and orchestrator/worker patterns.
• Build agents on Microsoft Copilot Studio and code-first frameworks (Semantic Kernel, LangGraph, LangChain, Autogen or equivalent).
• Engineer instructions, system prompts and reasoning flows; implement deterministic guardrails around non-deterministic model behavior.
• Define and implement tool/function calling - API actions, connectors, custom plugins - with robust error handling and fallback paths.
• Implement human-in-the-loop checkpoints, approval flows and escalation-to-human paths where decisions carry business or regulatory risk.
• Build agent memory and state management - conversation context, session state and long-term knowledge persistence.
• Build RAG pipelines: chunking strategy, embedding selection, vector indexing, hybrid and semantic retrieval, re-ranking.
• Connect agents to enterprise knowledge sources - SharePoint, Dataverse, Graph connectors, relational stores and document repositories.
• Tune retrieval quality and diagnose grounding failures, hallucination and citation accuracy issues.
• Implement source attribution so every agent answer is traceable to its grounding evidence.

Skills

Python

C#/.NET

API design

LLM frameworks

Prompt engineering

Agent orchestration

RAG pipelines

Vector stores

Grounding attribution

Tools

Semantic Kernel

LangGraph

LangChain

Autogen

Azure AI Search

Chroma

pgvector

Job description

Design, build, and deploy enterprise-grade AI agents using code-first frameworks, with strong focus on agent architecture, tool integration, grounding/RAG, and secure identity patterns. Ensure agents are production-ready through measurable evaluation, traceable source attribution, human-in-the-loop controls, CI/CD deployment, and stakeholder-driven delivery.

Key Responsibilities

Agent Design & Build

• Design agent architectures - single-agent, multi-agent and orchestrator/worker patterns - selecting the right pattern for each use case.
• Build agents on Microsoft Copilot Studio and code-first frameworks (Semantic Kernel, LangGraph, LangChain, Autogen or equivalent).
• Engineer instructions, system prompts and reasoning flows; implement deterministic guardrails around non-deterministic model behavior.
• Define and implement tool/function calling - API actions, connectors, custom plugins - with robust error handling and fallback paths.
• Implement human-in-the-loop checkpoints, approval flows and escalation-to-human paths where decisions carry business or regulatory risk.
• Build agent memory and state management - conversation context, session state and long-term knowledge persistence.
• Build RAG pipelines: chunking strategy, embedding selection, vector indexing, hybrid and semantic retrieval, re-ranking.
• Connect agents to enterprise knowledge sources - SharePoint, Dataverse, Graph connectors, relational stores and document repositories.
• Tune retrieval quality and diagnose grounding failures, hallucination and citation accuracy issues.
• Implement source attribution so every agent answer is traceable to its grounding evidence.

Integration & Deployment

• Integrate agents with enterprise systems - ServiceNow, Microsoft Graph, Power Platform, REST APIs and internal services.
• Implement authentication and authorization patterns - Entra ID app registration, SSO, managed identity, delegated vs application permissions, least privilege.
• Deploy agents through Teams, M365 Copilot, web channels and API endpoints as the use case requires.
• Build CI/CD pipelines for agent solutions with environment promotion (dev → test → prod) and version control.

Required Skills & Experience

• 5+ years of relevant professional experience.
• Demonstrable production delivery of at least two LLM-based or agentic applications - not prototypes or demos.
• Strong Python (preferred) and/or C#/.NET, with clean API design and solid software engineering fundamentals.
• Hands‑on with an agent framework: Semantic Kernel, LangGraph/LangChain, Autogen, or equivalent.
• Deep prompt engineering skill: system prompt design, few‑shot patterns, structured output, chain-of-thought control, instruction hierarchies.
• RAG implementation experience end to end, including vector stores (Azure AI Search, Chroma, pgvector or similar) and retrieval tuning.
• Tool/function calling design including schema definition, argument validation and failure recovery.
• Enterprise identity and API integration - Entra ID, OAuth 2.0, app registrations, scoped permissions.
• Evaluation discipline - able to prove an agent works with evidence, not anecdote.
• Excellent stakeholder communication; comfortable sitting with a business SME and turning a conversation into a working agent.

Preferred

• Experience in financial services, insurance or another regulated industry with formal AI governance gates.
• Multi‑agent orchestration at production scale.
• Fine‑tuning, distillation or small‑language‑model deployment experience.
• Familiarity with agent cost management and chargeback/tokenomics modelling.