Role: Lead – Principal Agentic AI Engineer
Level: Lead – Principal (Onsite Technical Lead)
Leadership Scope: Leads an onsite and offshore Agentic AI engineering pod and serves as the primary technical contact for the client

About the Role

We are looking for a Lead / Principal Agentic AI Engineer to serve as the senior-most Agentic AI engineer onsite, leading the technical relationship with an enterprise client and driving the delivery of production-grade Agentic AI platforms that automate document-intensive, decision-heavy workflows.

You will define the architecture, present and defend technical decisions with the client’s enterprise architects and security and compliance teams, and lead a blended onsite and offshore engineering pod through successful delivery.

You will operate in highly constrained enterprise environments involving self-hosted models for PII-sensitive workloads, restricted external inference, controlled MCP integrations, immutable audit trails for every agent action, and aggressive delivery timelines. You will be responsible for designing architectures that perform reliably within these constraints while keeping the engineering team aligned and moving forward.

What You’ll Drive

• Own the end-to-end architecture of Agentic AI document intelligence platforms, including the retrieval layer, RAG and extraction engines, abstracted consumption APIs, and multi-agent orchestration, designed as decoupled and independently deployable services.
• Define the MCP connector framework and source onboarding strategy using a configuration-driven model that enables new internal systems and external sources to be added through connector types, authentication profiles, and extraction templates without requiring changes to the core codebase.
• Drive model governance by working within the client’s approved and self-hosted model ecosystem and designing guardrails, evaluation frameworks, confidence thresholds, and human-in-the-loop routing for high-risk decisions.
• Provide technical leadership to the onsite and offshore engineering pod through design reviews, code quality standards, delivery sequencing, mentoring, and issue resolution while remaining hands-on with the most complex areas of the codebase.
• Own the client-facing technical narrative by translating requirements into architecture, articulating and defending design decisions, and managing scope against fixed delivery timelines.
• Ensure non-functional rigor, including horizontal scaling on OpenShift, circuit breaker and retry mechanisms for external integrations, OAuth 2.0/RBAC enforcement, and comprehensive OpenTelemetry and Datadog observability.

Must-Have Skills

• 8–10 years of software engineering experience, including 3+ years architecting and delivering LLM, GenAI, or Agentic AI systems in production.
• Deep RAG expertise, including designing retrieval systems at scale, chunking and indexing strategies, embedding model selection, hybrid and semantic retrieval, re-ranking, evaluation, and troubleshooting common failure modes.
• Strong MCP and connector experience, with proven expertise in designing tool-calling and connector frameworks and defining reusable, configuration-driven onboarding patterns for heterogeneous internal and external data sources.
• Expertise in multi-agent orchestration using LangGraph, Google ADK, or equivalent frameworks, with a strong understanding of agent topology, state management, retries, determinism, and when not to use an agent-based approach.
• Strong Python and FastAPI expertise for production services, along with a solid understanding of Java/Spring Boot and PostgreSQL to lead integrations across a polyglot technology stack.
• Strong knowledge of LLM evaluation, guardrails, prompt strategy, and hallucination control as core engineering disciplines.
• Demonstrated technical leadership experience, including leading engineering pods, establishing standards, mentoring senior engineers, and managing client and stakeholder relationships.
• Experience delivering solutions in regulated or security-constrained environments, with a strong understanding of audit requirements, PII protection, SSO/RBAC, and model governance.
• Excellent executive-level communication skills, with the ability to serve as the technical voice in client discussions.

Key Skills

Agentic AI/LLM, RAG, MCP & Multi-Agent Orchestration, Python/FastAPI, LangGraph/Google ADK, LLM Evaluation & Guardrails, Vector Databases, Document AI, and Enterprise AI Architecture.

Good to Have

• Strong ML modeling knowledge, including fine-tuning and adapter training, embedding model evaluation and selection, classical ML for classification and confidence scoring, and the development of rigorous evaluation frameworks.
• Experience with self-hosted or on-premises LLM serving, including inference optimization, GPU-aware deployment, and PII-safe inference.
• Document AI / Intelligent Document Processing (IDP) expertise, including layout-aware extraction, OCR pipelines, and extracting information from footnotes, tables, and complex forms.
• Experience with OpenShift/Kubernetes architecture, OpenTelemetry, Datadog, and enterprise secrets management platforms such as HashiCorp Vault or CyberArk.
• Experience with vector database architecture and performance tuning at scale using Qdrant, pgvector, or similar technologies.
• Prior onsite delivery or client-facing technical leadership experience in the U.S.

Why This Role

You will serve as the technical lead for high-stakes Agentic AI initiatives in demanding enterprise environments, owning the solution architecture, leading the engineering team, and delivering production systems that operations and compliance teams rely on every day.

This role offers a high degree of autonomy, direct client influence, and the opportunity to shape and deliver enterprise-grade Agentic AI solutions.

The Stack You’ll Work With

• Agent & Orchestration: Python, FastAPI, multi-agent orchestration frameworks (LangGraph, Google ADK, or equivalent), MCP (Model Context Protocol) connectors
• RAG & Extraction: Embeddings, vector stores (Qdrant, pgvector), chunking and retrieval strategies, OCR and GenAI extraction pipelines, confidence scoring, and structured-output schemas
• Models: Client-approved, self-hosted LLMs for PII-sensitive workloads, prompt engineering, evaluation, and guardrails
• Backend: Java, Spring Boot microservices, PostgreSQL, REST APIs (synchronous and asynchronous)
• Frontend: React 19, TypeScript
• Platform: OpenShift, OAuth 2.0, Okta SSO, Active Directory-based RBAC, OpenTelemetry, Datadog, and immutable audit logging