Research Engineer — ML / RL / Agents at Affine

About Affine

Affine is building an incentivized reinforcement learning (RL) environment that pays miners for measurable improvements on complex tasks like program synthesis and reasoning.

Running on Bittensor’s Subnet 120, Affine has created a sybil-proof, decoy-proof, copy-proof, and overfitting-proof mechanism that rewards genuine innovation in model performance.

Our long-term goal is to commoditize reasoning—the highest form of intelligence—by coordinating a large, permissionless network of contributors working on RL tasks that collectively push the limits of open, distributed AI.

The Role

We’re looking for research-minded engineers who thrive at the intersection of machine learning, reinforcement learning, and systems engineering.
Your work will focus on experimentation, rapid prototyping, and discovery: designing post-training methods, exploring agent architectures, and validating new ideas in live, competitive benchmarks.

You’ll take cutting-edge theory—GRPO, PPO, multi-objective RL, program abduction—and turn it into working systems that miners can train, evaluate, and monetize through Affine on Bittensor’s Subnet 120.

This role is ideal for someone who moves seamlessly between research and production, capable of translating papers into reproducible code and scaling new ideas into pipelines that power a decentralized, incentive-driven RL ecosystem.

Responsibilities

• Design decentralized RL systems that incentivize miners to train, refine, and host high-quality agentic LLMs on the Bittensor subnet.

• Build evaluation frameworks to measure model performance, safety, and alignment—including task design, metrics, adversarial testing, and red-teaming.

• Research and apply cutting-edge RL and alignment techniques to strengthen the training–evaluation feedback loop.

• Prototype and scale algorithms: explore new agent architectures and post-training strategies, then operationalize them into reproducible pipelines.

• Contribute to live competitive benchmarks, ensuring systems reward genuine reasoning improvements instead of gaming behaviors.

Requirements & Qualifications

• Proven expertise in reinforcement learning, with hands-on experience designing and tuning RL algorithms.

• Strong engineering skills in Python and experience with PyTorch, JAX, or TensorFlow.

• Experience with distributed systems and scaling high-performance ML infrastructure.

• Familiarity with LLMs and tool-use patterns, including APIs, external integrations, and function calling.

• Advanced academic or applied background—Master’s, PhD, or equivalent industry research experience.

• Bonus: Background in multi-agent systems, mechanism design, or RLHF.

Bonus Qualifications

• Publications in top AI/ML venues (NeurIPS, ICML, ICLR, AAAI) in RL, game theory, AI safety, or decentralized AI.

• Experience with virtualization and sandboxed execution for safe agent tool use.

• Knowledge of game theory and advanced mechanism design.

• Contributions to major open-source RL or LLM projects.

Why Affine

• Operate at the frontier of decentralized AI, merging reinforcement learning, open-source collaboration, and crypto incentives.

• Collaborate with leading minds in AI research, incentive design, and distributed systems.

• Help define how reasoning agents are trained, evaluated, and rewarded in open networks.

• Join a lean, high-trust team working to rethink how intelligence evolves online.