AI Product Operator

This is a fully remote position, open to applicants in Nepal.

📋 Description

• Create sandboxed, stateful environments that encompass real Niural payroll, contractor onboarding, invoice reconciliation, and compliance workflows.

• Ensure environments are seeded and replayable, incorporating difficulty controls and structured curricula.

• Produce synthetic data for these environments, as actual customer payroll data must remain within Niural systems.

• Develop programmatic verifiers based on established ground truth and reduce reliance on model-as-judge techniques.

• Evaluate trajectories and protect against reward hacking and contamination.

• Execute reinforcement learning loops, curriculum scheduling, fine-tuning, and distillation as appropriate.

• Set baselines, conduct ablations, and assess if improvements generalize effectively.

• Document and publish internal negative results that are measured.

• Prototype agent architectures, memory strategies, tool-use patterns, and scaffolds within one- to two-week development spikes.

• Transition successful prototypes to production-agent teams and maintain involvement throughout deployment.

• Authorship of first-author research papers for Niural AI, ensuring reproducible methodologies and candid limitations.

• Generate internal technical reports and represent work through preprints, presentations, and potentially open-sourced environments.

• Collaborate within the product team alongside the AI Product Manager and production-agent engineers.

• Conduct experiments that significantly impact shipped products and customer-facing systems.

⛳️ Requirements

• A published, peer-reviewed paper presented at a conference or in a journal; this is a non-negotiable requirement.

• Fundamental understanding of AI and machine learning principles.

• Proficiency in Python, particularly with numpy and the scientific stack.

• Capability to write code that is conducive for others to build upon.

• Experience in creating and managing services, including APIs, databases, queues, Docker, deployment, logging, and debugging.

• Practical experience with agent systems, harnesses, tool usage, and multi-step workflows that operate autonomously.

• Familiarity with reinforcement learning; ability to interpret and implement policy optimization research.

• Knowledge and opinions on reward design methodologies.

• Commitment to reproducibility; enabling others to clone the repository, follow the README, and replicate results.

• Demonstrated evidence of impactful work, such as a utilized product, cited or extended research, open-source projects with active users, or systems that influenced corporate decisions.

• Willingness and ability to work a permanent night shift aligned with US hours.

• Hands-on experience with gym-style environments or reinforcement learning training frameworks.

• Expertise in evaluation tooling, harness engineering, or benchmark development.

• Experience in synthetic data generation, fine-tuning, or distillation.

• Knowledge of Docker, sandboxing, and distributed training infrastructure.

• Contributions to open-source projects that are relied upon by others.

• Familiarity with fintech, payroll, HR tech, or compliance sectors.

• Experience in writing beyond academic papers, such as blogs, documentation, or teaching materials.

🏝️ Benefits

• Competitive salary package tailored to acknowledge your expertise and contributions.

• Access to ongoing learning and career development programs.

• Opportunities for mentorship and coaching to facilitate your career growth.

• A supportive and collaborative work environment that values ideas and encourages teamwork.

• Complimentary snacks and lunch provided to maintain your energy throughout the shift.

• Provision of the latest MacBook Pro and a high-performance monitor.

• "Fast track" admission to selected US universities for outstanding performers interested in pursuing a Master's in CS in the US.

• Sponsorship for L1 visas and immigrant visas (Green Card) for high performers wishing to move to the US.

• Opportunity to publish research under your name.

• Share in the financial benefits of building foundational internet infrastructure.