Job Description
About Abundant
Abundant builds reinforcement learning environments for frontier AI labs. We design and operate the simulation infrastructure where next-generation models learn to reason, act, and solve complex problems. We’re a small, high-impact team scaling fast.
The Role
Research PM (RPM) is an emerging role that we are pioneering at Abundant. RPMs are PMs of the model, unlike traditional PMs who own an application or a feature. RPMs design and implement model capabilities, either as part of an AI lab, data lab, or other research organization.
Like traditional PMs, RPMs are still the voice of the user. RPMs talk to users or enterprises to find out how their work is performed and how they use AI. Then they decide how to distill the users’ feedback into evaluation and training data. We are RPMs that are triple-threats: folks that read up on the latest machine learning literature, and can and own technical tooling, and can scale up teams of hundreds. For folks in the AI data industry, you’ll recognize that this is the evolution of the SPL role from an purely-operations job into an all-rounder that includes research, product and execution.
An RPM will end-to-end execution of Abundant’s most critical projects. You will translate ambiguous research requirements into structured workflows, manage distributed expert teams, and deliver flawless results under tight timelines. This is a founding-level operations role with outsized ownership.
What You’ll Do
- Own project delivery end-to-end — scoping, design, contributor ops, QA, and handoff to AI lab partners
- Build and manage relationships with researchers at frontier labs
- Recruit, scale, and run distributed teams of domain experts — sourcing, performance, quality, engagement
- Build the operational playbook — processes, tooling, and QA systems that don’t exist yet
Who you are
- Relentless executor — you have extreme agency and drive; when you see a problem you fix it, when a path doesn’t exist you create one
- Thrive in chaos — ambiguity energizes you; you’ve built systems and processes from nothing in fast-moving environments
- High-autonomy track record — experience in operations, consulting, or program management where you owned outcomes, not just tasks
- Technically fluent — you understand software systems and data pipelines well enough to earn credibility with engineers
Nice to Have
- Familiarity with AI evaluation, benchmarks, or reinforcement learning
Requirements
Deep experience in one or more of the following:
- ML engineering or research
- Reinforcement learning and/or agentic harnesses
- Agent evaluation and benchmarking
- High-scale and high-reliability batch data pipelines
Optimize Your Resume for This Job
Get a match score and see exactly which keywords you're missing
Job Details
- Category
- Research
- Employment Type
- Full Time
- Location
- San Francisco, CA
- Posted
- Compensation
- $120,000 - $285,000 per year
About Abundant
Hello! 👋 We are a team of former ML engineers, founders, roboticists and ops leads who obsess about data and its impact on safe, reliable AI. We specialize in creating environments and datasets for RL by leveraging our experience in simulation and model training. By the numbers: • Powering 3 of the top 6 global AI labs and multiple Fortune 500 enterprises • Billions of training tokens generated each month, 2x month over month • Exclusive, global network of over 500 domain experts We believe humans are inherently creative, and thrive by pushing the frontier. We are working towards an abundant future--one where everyone has access to infinite intelligence, services and goods. Based in San Francisco, CA. We enjoy good food and good company. -- more info below -- Abundant is building the NVIDIA of training data. AI models rely on two fundamental ingredients: compute and data. NVIDIA, the leader in compute, has a peak market cap of $5T and generated $130B in revenue last year as the need for scaling compute has exploded. We believe the need to scale data is just beginning, as we move beyond SFT and human supervision to RL and Learning from Experience. Our founding team consists of second-time founders, ML engineers and data leads from Waymo, Google, Meta and AWS. Our team has previously collaborated with DeepMind to classify hate speech in YouTube videos, trained SOTA models for self-driving, and scaled data pipelines with thousands of human annotators. Our pioneering work in human computation, synthetic data, imitation learning and RL give us a solid advantage in delivering results to our customers. Why now? Training data is more important and more scarce than ever before. Scaling laws dictate that linear improvement in model performance demands an exponential increase in training data. But there is only one World Wide Web and most of it has already been trained on. The next advances will require new, diverse, and high-quality datasets, making training data more important and scarce than ever before. What happens if we succeed? Abundant will be the core enabler for AGI and beyond. Most of the challenges in model training are already solved. What’s missing is the data necessary to move from general knowledge to domain expertise; from chatbots to agents; and from digital intelligence to physical AI. Ask any AI researcher or roboticist: the core bottleneck to progress is the availability of data, i.e. “abundant data”. Abundant works with the most advanced AI labs and startups, as well as F500 enterprises.
More Roles at Abundant
Similar Research Roles
Found this role interesting?