LLM Engineer
Summary
Founding LLM engineer at an early-stage, YC-backed AI supply chain startup, building and owning LLM-powered agent systems end-to-end that automate production, warehousing, and distribution decisions.
About the Role
You'll be a founding LLM engineer at an early-stage, Y Combinator–backed AI supply chain startup, building an AI Supply Chain Manager from the ground up. This is a high-autonomy role where you'll own LLM systems end-to-end—from prototype to production—turning the physical flow of goods into something as programmable as code.
What You'll Do
Build and iterate on LLM-powered agents that decide what to produce, where to make it, and how to move it through factories, warehouses, and distribution channels.
Design robust prompts, tool definitions, structured outputs, and multi-step agent flows that handle edge cases in messy real-world data.
Select and evaluate LLMs based on latency, cost, accuracy, and use-case fit across supply chain workflows.
Ship LLM features from concept to production, owning the full pipeline from API integration to tested, evaluated systems.
Build and maintain evaluation frameworks to iterate quickly and safely on prompts and agent behaviors.
Set up observability, monitoring, and feedback loops to track production accuracy and continuously improve system performance.
What We're Looking For
3+ years of hands-on experience shipping LLM-powered features or systems in production environments.
Practical experience with modern LLM APIs (OpenAI, Anthropic, DeepSeek, Gemini, or similar) and a track record of shipping real features with them.
Experience building LLM-powered agents or automations that are core to production systems—not just internal demos.
Strong system-design skills covering tool/function design, structured outputs, and multi-step agent flow architecture.
Proficiency designing and maintaining evaluation pipelines for prompts, workflows, and LLM outputs.
Experience with LLM observability, monitoring, and logging tools to track production accuracy, latency, and quality metrics.
Demonstrated ability to evaluate and select LLMs based on latency, cost, capability, and use-case tradeoffs.
Hands-on experience with the HuggingFace ecosystem in real production or prototype projects.
Bonus: Experience running self-hosted LLMs in production.
Bonus: Background in supply chain, logistics, or operations domain systems.
Compensation & Benefits
Visa sponsorship is available.
Location
On-site in New York, NY, United States.
As published by ashby
Name, Email, Resume
- Do you have work authorization to work in that country? yes / no