About the role
About Hark
Hark is an artificial intelligence company building advanced, personalized intelligence. One that is proactive, multimodal, and capable of interacting with the world through speech, text, vision, and persistent memory.
We're pairing that intelligence with next-generation hardware to create a universal interface between humans and machines. While today's AI largely operates through chat boxes and decade-old devices, Hark is focused on what comes next: agentic systems that interact naturally with people and the real world.
To get there, we're developing multimodal models and next-generation AI hardware together - designed from the ground up as a single, unified interface for a new era of intelligent systems.
About the Role
At Hark, you'll lead the development for the foundation of agentic reinforcement learning (RL) environments: the digital worlds in which AI agents learn to use software, navigate complex workflows, and act on their surroundings. Our ambition is to simulate the breadth and complexity of the internet, connecting websites, apps, services, and artifacts into rich, stateful environments, with the potential to evolve dynamically over time.
This is a chance to shape both the worlds agents learn in and the systems that make learning at scale possible. You'll build environments end-to-end and work closely with researchers to bring them into RL training and evaluation. You'll also develop the infrastructure behind our agent gym, making these environments easy to integrate, reproduce, and scale across heterogeneous execution targets.
Responsibilities
- Build agent environments and simulations end-to-end, including frontend interfaces, backend services, APIs, data models, tools, and realistic workflows used to train and evaluate AI agents.
- Build the infrastructure that powers our agent gym, including orchestration, sandboxing, packaging, benchmarking, and systems for running environments across heterogeneous targets.
- Create interconnected simulations with coherent behavior and shared state across websites, apps, and artifacts. Explore dynamic environments with changing content, background activity, and events over time.
- Partner with researchers to integrate environments into RL training and evaluation pipelines, building reusable agent interfaces, environment lifecycle APIs, and reliable collection of trajectories and artifacts.
- Build infrastructure that supports thousands of concurrent sandboxed environments, with isolation between episodes, consistent state, snapshots and resets, and recovery from partial failures.
- Improve startup and teardown latency, scheduling, resource utilization, and throughput to make large-scale RL rollouts efficient and cost-effective.
- Build observability, debugging, and reproducibility tools. Diagnose resource contention, timeouts, and runtime failures, and distinguish infrastructure problems from agent behavior.
Requirements
- Strong full-stack engineering skills, with experience building applications across frontend interfaces, backend services, APIs, and data models.
- Experience with stateful backend or distributed systems, including databases, queues, caching, concurrency, consistency, and failure recovery.
- Strong algorithms and systems fundamentals. You can profile complex workloads, identify bottlenecks, and validate performance improvements without compromising correctness.
- Ability to design clear, reusable APIs and abstractions that make complex environments easy for researchers and engineers to integrate, extend, and operate.
- Careful debugging and testing habits, with attention to application fidelity, state transitions, reproducibility, and failures that can silently corrupt results.
- Creativity and ownership in ambiguous problems. You can turn real software behavior into working simulations, collaborate closely with researchers, and learn the agent and RL systems your environments support.
Bonus Qualifications
- Experience integrating environments with RL rollout systems, agent training pipelines, or evaluation harnesses.
- Experience with containers, virtual machines, sandbox runtimes, or orchestration systems for large-scale distributed workloads.
- Experience with browser automation, computer-use agents, simulation frameworks, or dynamic systems driven by events.
Compensation
The US base salary range for this full-time position is between $170,000 - $400,000 annually.
The pay offered for this position may vary based on several individual factors, including job-related knowledge, skills, and experience. The total compensation package may also include additional components and benefits depending on the specific role. This information will be shared if an employment offer is extended.