A

Abundant

On-demand human operators ensure AI reliability

Member of Technical Staff - Research

AI ResearchConfirmed live in the last 24 hours
Full-timeSan Francisco, CA, USA$250k - $450kIn-person

About the Job

Requirements

  • Write substantial code and demonstrate strong software engineering skills and deep Python proficiency, including end-to-end system ownership and debugging unfamiliar layers.
  • Have built evaluations, reinforcement learning environments, agent harnesses, verifiers, or execution sandboxes, and be able to assess their trustworthiness.
  • Have shipped research into production, particularly post-training, distillation, and high-stakes evaluation.
  • Have worked on large-scale agent systems, including orchestration frameworks, tool APIs, distributed execution, observability, and logging.
  • Reach ground truth in unfamiliar domains by building and examining outputs.
  • Identify misspecified tasks, ambiguous prompts, and graders that can be gamed.
  • Have a thoughtful perspective on the societal and safety impacts of deploying general-purpose AI systems.

Responsibilities

  • Architect and execute a research agenda to find generalizable ideas that advance model reasoning at scale, carrying work from hypothesis through live system training runs.
  • Build long-horizon agentic tasks that remain difficult across model generations, with graders that award meaningful partial credit and resist obvious exploits.
  • Analyze model trajectories to diagnose failure modes and develop the data or reward signals that address them.
  • Partner with advanced AI research teams on datasets and benchmarks shaping frontier-model behavior, with a focus on alignment, safety, and reward-signal design.
  • Identify, scope, and run long projects, selecting problems important to scaling data toward AGI.
  • Work with engineering on data pipelines, internal tooling, and deep learning implementations, and fix workflows that produce poor data.

Desired Qualifications

  • A PhD in Computer Science, machine learning, or natural language processing, and publications at NeurIPS, ICML, or ICLR; neither is required.
  • Work on public benchmarks or task suites such as Terminal-Bench, SWE-bench-style harnesses, or HCAST.
  • Reinforcement learning from human feedback or reinforcement learning from AI feedback experience, with a perspective on how training choices affect model behavior.
  • Experience running work through large expert or annotation operations.

About the Company

Abundant

On-demand human operators ensure AI reliability

Abundant provides a human-in-the-loop platform that makes AI agents reliable in production, especially for healthcare, legal, and financial services. Its system runs AI tasks and, when an edge case arises, calls a Human Intelligence API where expert operators complete the task and hand control back; all interactions are logged to create labeled training data and improve models, with an analytics dashboard for monitoring. The key difference is a dedicated human-override layer and a focus on real-world reliability in regulated industries, backed by founders with extensive experience in planning systems, content classification, and cloud infrastructure. The goal is to increase AI accuracy and task success in production by combining automated agents with precise human supervision and data-driven model improvement.

Company size
11-50
Company stage
Seed
Total funding
$130K
Headquarters
San Francisco, CA
Founded
2024

Recently Posted Jobs

Explore more opportunities at Abundant

See more jobs