Senior - AI Engineer, Development (51505)

Citrin Cooperman offers a dynamic work environment, fostering professional growth and collaboration. We’re continuously seeking talented individuals who bring a problem-solving mindset, fresh perspectives, and sharp technical expertise. We know you have choices, so our team of collaborative, innovative professionals are ready to support your professional development. At Citrin Cooperman, we offer competitive compensation and benefits and most importantly, the flexibility to manage your personal and professional life to focus on what matters most to you!

We are seeking a Senior – AI Engineer, Development, to join our Development team within the Information Technology department. The AI Solutions team is the vanguard of our enterprise AI competency, bridging the gap between rapid, external SI-led pilot programs and enterprise operations. They’ll serve as the first line of defense and primary code owner for all inbound agentic AI applications.

In this highly-on role, you’ll be the first internal engineer to receive, review, and evaluate the AI codebase delivered by our System Integrator Partners. You’ll take non-deterministic, proof-of-concept agent workflows (built in frameworks like LangGraph or Semantic Kernel), and rigorously refactor them for a highly regulated enterprise environment. You’ll pair closely with our Senior Cloud and Data Engineers to ensure the agents are seamlessly integrated within our core infrastructure, including but not limited to secure connections to our data resources in the Microsoft Fabric ecosystem. The ideal candidate has likely spent time on both sides of the fence, perhaps building rapid prototypes in a strategic innovation or consulting capacity, and now operating within a complex enterprise, as well as architecting robust Python solutions that scale securely.

Responsibilities are, but not limited to

• Codebase Reception & Refactoring: Take ownership of SI-developed agentic codebases. Conduct rigorous code reviews, eliminate any technical debt, and refactor prompt chains, as needed, into resilient, modular code ready for production workloads.
• Agentic State & Memory Management: Industrialize how agents handle state, memory, and reasoning. Implement robust error handling, dynamic retry logic, and fallback routing for complex workflows (e.g., standardizing LangGraph state transitions).
• Cross-Functional Pairing: Work daily alongside the Senior Data Engineer to ensure the agents are querying high-quality, formatted data, and pair with the Senior Cloud Engineer to package and deploy the agents within secure, private network boundaries.
• Evaluation Implementation: Translate the AI Product Manager’s “value metrics” into actual code. Instrument the codebase with telemetry to capture reasoning traces, latency, token consumption, and precision/recall metrics for continuous LLMOps monitoring.
• Security & Guardrail Enforcement: Hardcode enterprise security guardrails directly into the agent’s logic. Ensure tools and APIs invoked by the LLM strictly adhere to internal IAM policies and data access boundaries.
• Performance Optimization: Optimize the codebase for production latency. Refine context window management, optimize embedding retrieval, and adjust chunking strategies to ensure the agent responds quickly and cost-effectively.

The ideal candidate must:

• Have a bachelor’s degree in computer science, information technology, engineering, or equivalent practical experience.
• Be Microsoft Certified: Azure AI Engineer Associate (AI-102)
• Have 5+ years of software engineering experience with deep, production-level expertise in Python.
• Have significant hands-on experience building generative AI applications and multi-agent workflows using frameworks such as LangChain, LangGraph, or Semantic Kernel.
• Have proven experience operating in complex, large-scale enterprise environments.
• Be comfortable interfacing with modern data platforms (e.g., Databricks, Microsoft Fabric) and transition code from rapid prototyping into hardened operations.
• Have experience with LLMOps principles, including prompt versioning, continuous evaluation (e.g., LangSmith, Braintrust), and tracing non-deterministic outputs.
• Have a demonstrated ability to conduct rigorous code reviews and refactor external or legacy codebases.
• Have a strong understanding of RAG (Retrieval-Augmented Generation) architectures, vector databases, and the API ecosystems of frontier models (Anthropic, OpenAI, Google.
• Be the “Hacker-to-Engineer” Bridge: Appreciate the creativity of a fast, duct-taped SI prototype, but possesses the uncompromising discipline to rewrite it for high-availability enterprise scale.
• Be radically Collaborative: Naturally default to pairing with Data and Cloud operations rather than working in an isolated AI silo.
• Be a pragmatic Problem Solver: Comfortable with the inherent “weirdness” of LLMs. Does not expect perfect determinism but knows exactly how to build software guardrails to contain and manage probabilistic behavior.