Research Engineer, Agentic Systems
"About Ramp Labs"
About This Role
Ramp Labs is Ramp's AI research and engineering group. The team post-trains models on finance work, builds the environments and evaluations those models learn from, and ships them into the products used by Ramp's 85,000+ customers.
What You'll Be Doing
Deciding how work gets broken down, what context each agent receives, how many agents to run, and which model handles each step
Improving accuracy through context design, self-verification, and voting/consensus across agents, while recognizing when adding more agents stops helping
Measuring quality without reviewing every trace, using sampling, automated verification, and precision/recall on final outputs
Reducing cost and latency per useful result through model routing, caching, early stopping, and smaller or fine-tuned models where appropriate
Converting one-off experiments into reusable workflows on the existing orchestration stack
Evaluating new models, inference setups, and training approaches, and incorporating what proves effective
What We're Looking For
Experience building applied AI systems and LLM applications that have shipped
Hands-on experience with multi-agent or high-fan-out LLM workloads, such as parallel agents, map-reduce over models, or large batch inference
Strong Python engineering skills, with the ability to build concurrent, distributed systems that hold up in production
Judgment about desired outputs-defining what the end result should look like before scaling up agent swarms, and distinguishing signal from noise
A habit of measuring cost, yield, false-positive rate, and whether changes actually help
Ownership of a roadmap and production system with minimal oversight
Clear written communication
Bonus Points
Experience building agent orchestration or job systems (such as Ray, Temporal, Modal, or custom solutions)
Experience using LLMs to search large spaces, including codebases, document sets, logs, or other large corpora
Experience post-training or hosting open-source models for high-volume inference
Background in distributed systems
Public work that demonstrates capability, such as an agent framework, open-source project, write-up, or paper
Ready to Apply?
To submit your application for this position, please visit Ramp Network's official website and follow their application process.