Senior AI Agentic Engineer

BDIPlus•New York, NY

About The Position

BDIPlus is seeking a Senior AI Agentic Engineer to support a Fortune 100 financial services client’s enterprise data products and real-time intelligence initiatives. In this role, you will design, build, and deploy production-grade AI agents using Claude. These agents will deliver real-time intelligence within a highly governed ecosystem powered by MCP tool integration. You’ll work at the forefront of agentic AI in a regulated environment, ensuring performance, trust, and compliance at scale.

Requirements

3+ years of experience building production AI/ML applications, including at least 1 year working with LLMs
Hands-on experience with Claude or equivalent platforms (Azure OpenAI, SageMaker)
Strong understanding of Model Context Protocol (MCP), tool-use patterns, and agent-based architectures
Experience configuring API gateways (e.g., Kong Gateway, AWS API Gateway) including authentication and rate limiting
Proficiency with observability and audit systems (ELK Stack or similar)
Experience with agent evaluation techniques: benchmarking, adversarial testing, prompt injection defense
Strong programming skills in Python (FastAPI preferred) or Node.js
Solid understanding of Responsible AI principles (fairness, explainability, governance, compliance)
Experience using AI-assisted development tools (Claude Code preferred; training available)

Nice To Haves

Experience with semantic layers and MCP integrations
Financial services or insurance domain experience
Familiarity with Elasticsearch ML or similar anomaly detection tools
Experience with data catalog and lineage tools
Experience with data quality platforms
Knowledge of regulatory frameworks (e.g., NAIC AI Bulletin, NY Reg 187, SOX)
Experience deploying AI agents with human-in-the-loop governance in regulated environments

Responsibilities

Design and develop production AI agents
Build agents on Claude with scalable, secure deployment patterns
Configure and manage reverse proxies / gateways with OIDC / Rate limitations etc.
Implement comprehensive ELK-based audit trails with long-term retention (including S3 Glacier archival)
Design human-in-the-loop workflows with approval gates, task queues, and escalation rules—especially for actuarial outputs
Conduct accuracy benchmarking, adversarial testing, and Responsible AI evaluations aligned with regulatory expectations (e.g., NAIC AI guidance)
Implement role-based and attribute-based access controls (RBAC/ABAC) via MCP and semantic layers (e.g., Denodo)
Integrate data quality signals into agent responses
Build safety mechanisms such as circuit breakers, usage limits, and error threshold controls
Collaborate with streaming/data engineering teams to connect agents to real-time data pipelines (Kafka, Iceberg)
Leverage AI-assisted development tools (Claude Code) to accelerate delivery