AI/ML Engineer

AppZen is the leader in autonomous spend-to-pay software. Its patented artificial intelligence accurately and efficiently processes information from thousands of data sources so that organizations can better understand enterprise spend at scale to make smarter business decisions. It seamlessly integrates with existing accounts payable, expense, and card workflows to read, understand, and make real-time decisions based on your unique spend profile, leading to faster processing times and fewer instances of fraud or wasteful spend. Global enterprises, including one-third of the Fortune 500, use AppZen’s invoice, expense, and card transaction solutions to replace manual finance processes and accelerate the speed and agility of their businesses. To learn more, visit us at www.appzen.com.

About the role :

We are looking for an AI/ML Engineer with 1–3 years of experience to join our growing AI team. In this role, you will build and ship the LLM-powered and agentic features that read, understand and make decisions on invoices, expense reports and card transactions for some of the world’s largest finance teams. You will work alongside senior AI/ML engineers, data scientists and product managers to take features from prototype to production, and own them once they are live. If you enjoy turning the latest advances in LLMs and agents into reliable and measurable products, AppZen is the place for you.

Key Responsibilities :

• LLM & Agent Development: Build and improve LLM-powered features and multi-step agentic workflows (e.g., using LangGraph) that extract, validate and reason over invoices, receipts, expense reports and card transactions.

• Retrieval-Augmented Generation (RAG): Build retrieval pipelines (chunking, embeddings, vector search and re-ranking) that give models the right policy, vendor and transaction context to make accurate financial decisions.

• Tool Use & Structured Outputs: Implement function calling and API integrations so that models can reliably query internal services and ERP/AP systems, and return validated, structured (JSON) outputs.

• Evaluation & Quality: Build evaluation datasets, metrics and automated test suites to measure accuracy, hallucination rate, latency and cost, and use the results to improve prompts, retrieval and models.

• Production Engineering: Package and deploy models and services using Docker and CI/CD pipelines, add monitoring and logging, and keep them reliable, scalable and cost-efficient in production.