Applied Research Engineer

Nxt Level; San Francisco (CA); On-site; USD 180,000 - 280,000; Full time

Benefits offered by this job

Job summary

Nxt Level is seeking a Research Engineer / Scientist to own ambitious research bets end to end in San Francisco. You will drive hypothesis generation, data strategy, model training, evaluation, deployment, and iteration on reinforcement learning, post-training, and long-horizon autonomous systems.

The role welcomes fresh PhDs or senior researchers and offers the chance to shape product architecture while collaborating closely with founders and engineering leadership on technical strategy.

Qualifications

• Experience in frontier AI research, applied AI research, or research engineering.
• Strong background in ML, DL, RL, post-training, continual learning, or multimodal systems.
• Ability to design and run rigorous experiments.

Responsibilities

• Run experiments and train frontier AI models focused on human judgment scaling and intent representation learning.
• Post-train LLMs and multimodal agents using RL and continual learning methods.
• Work with large-scale screen recording and behavioral data to understand user workflows.

Skills

Frontier AI research

Reinforcement learning

Post-training learning

Continual learning

Multimodal systems

Experimental design

Programming fundamentals

Large-scale data handling

Education

PhD or MSc in Computer Science / ML

Tools

PyTorch

TensorFlow

Experiment tracking (Weights & Biases / MLflow)

Job description

Focus: Frontier AI, Agentic AI, Reinforcement Learning, Computer Use, Multimodal Models, Long-Horizon Agents

About Our Client

Our client is a stealth-stage AI company building frontier models focused on human intent understanding, computer use, and autonomous agent systems.

The founding team includes leaders from Tesla AI, Google DeepMind, NVIDIA, Physical Intelligence, and Apple. They are building from the ground up and looking for early research talent to help shape the technical direction, research agenda, and product architecture from day one.

This is an opportunity to work on some of the hardest problems in applied AI: teaching agents to understand how people work, represent intent, use computers, and complete long-horizon tasks with increasing autonomy.

About the Role

Our client is hiring a Research Engineer / Scientist to own ambitious research bets end to end.

This person will work across hypothesis generation, data strategy, model training, evaluation, deployment, and iteration. The role is focused on reinforcement learning, post-training, continual learning, multimodal agents, and long-horizon autonomous systems.

The team is open to a range of backgrounds, from fresh PhD graduates with strong research internships to senior researchers who have led teams at top AI labs.

What You’ll Do

• Run experiments and train frontier AI models focused on human judgment scaling, computer use, and intent representation learning
• Post-train LLMs and multimodal agents using reinforcement learning and continual learning methods
• Work with large-scale screen recording and behavioral data to understand how individual users work
• Build models that can learn user workflows and proactively automate tasks
• Design and execute rigorous research experiments that improve autonomous agent capabilities
• Own research bets end to end, from hypothesis and data through training, evaluation, deployment, and measurement
• Help define the company’s research direction as an early technical team member
• Partner closely with founders and engineering leadership on technical architecture and product strategy
• Translate cutting-edge research into systems that can power real product experiences

What We’re Looking For

• Experience in frontier AI research, applied AI research, or research engineering
• Strong background in machine learning, deep learning, reinforcement learning, post-training, continual learning, or multimodal systems
• Experience designing and running rigorous experiments
• Ability to move from research idea to implemented system
• Strong programming and engineering fundamentals
• Comfort working with large-scale datasets and complex model training workflows
• Strong judgment around model evaluation, data quality, experimental design, and deployment readiness
• Ability to operate in ambiguity and take ownership of open-ended research problems
• Clear communication and ability to collaborate with a small, high-caliber founding team
• Excitement about computer use, autonomous agents, human intent modeling, and the future of AI-native software

Relevant Research Areas

Relevant experience may include:

• Post-training
• Continual learning
• Long-horizon agents
• Human behavior modeling
• Intent representation learning
• Human judgment scaling

Ideal Background

Our client is open to a range of seniority levels, including:

• Fresh PhD graduates with strong research internship experience
• Research engineers from frontier AI labs
• Applied scientists with experience training and evaluating advanced AI systems
• Senior researchers who have led teams or major technical efforts
• Builders who can connect research quality with production-minded execution

Why This Opportunity

• Join a stealth-stage AI company at the founding team stage
• Work directly with leaders from Tesla AI, Google DeepMind, NVIDIA, Physical Intelligence, and Apple
• Help define the research agenda from day one
• Own ambitious technical bets across reinforcement learning, post-training, continual learning, and agentic AI
• Work on frontier problems in computer use, human intent understanding, and long-horizon autonomous agents
• Build models that learn how people work and help automate real tasks
• Shape both technical architecture and product direction early

Ideal Candidate Profile

The ideal candidate is a research-minded builder who wants to push the frontier of agentic AI.

They can design strong experiments, train models, reason deeply about evaluation, and turn ambiguous research questions into working systems. They are excited by the challenge of building AI agents that understand human intent, use computers effectively, and improve through real-world interaction.

This person wants to help shape the research foundation of a company from the earliest stage.