About AnalogyAI
We're building the data infrastructure behind self-evolving AI systems.
Our platform acts like an AI data researcher—automating the full lifecycle of data, from sourcing and curation to validation and delivery. Given a user prompt, it discovers and acquires relevant data sources, structures, and enriches them with task-specific schemas, runs rigorous quality and safety checks, and produces domain-specific, training-ready datasets and benchmarks.
Beyond data generation, we integrate directly with model training pipelines and production telemetry. By combining offline evaluations with real-world signals, our system automatically identifies failure modes and capability gaps, then generates the next most impactful dataset or incremental data batch to continuously improve the model.
We're building an inevitable data layer for enterprise AI—one that enables models to learn, adapt, and improve themselves over time.
About the Role
We're looking for a System-focused AI Researcher/Engineer to bridge the gap between low-level system control and high-level agent reasoning.
You won't just be prompting LLMs; you'll be building the environments they live in. Your mission is to solve the "Environment Bottleneck": how to record, replay, and scale agent interactions across thousands of virtualized OS instances. You will design the infrastructure for Computer Use Agents, architecting how we capture user trajectories and how we reliably "kick off" and manage complex environments at scale.
This is a "Research + Systems" role: you'll design the primitives for agent-OS interaction, validate them with SOTA models, and harden them into a production-grade data engine.
What You'll Do
- Architect Agent Environments: Build and optimize scalable environments for "Computer Use" agents (Windows/Linux/macOS virtualization, browser isolation, and containerized tool-use).
- System-Level Telemetry: Develop high-fidelity recording systems to capture user trajectories, UI metadata, and OS-level events to create gold-standard training data.
- Scalable Orchestration: Design and implement the infrastructure to "kick off" and manage thousands of concurrent agent environments for massive-scale data generation.
- Self-Evolving Loops: Work on the intersection of System + AI—using production failure signals to automatically spin up specific environment configurations for targeted data collection.
- Data Engineering for Agents: Build pipelines to process raw system traces into structured, fine-tuning-ready datasets (Reasoning chains, Action-Observation pairs).
- Benchmarking & Evaluation: Design robust "system-in-the-loop" benchmarks to measure agent reliability, safety, and recovery in real-world OS environments.
- Performance Optimization: Optimize the latency and throughput of agent-environment interactions to accelerate the research-to-production cycle.
What We're Looking For
- Strong Systems Background: Deep understanding of Linux/Unix/MacOS internals, virtualization (KVM/QEMU, Docker), or browser engines (Chromium/Playwright).
- Experience with Agentic Systems: Familiarity with LLM-based OS control, UI automation, or building "Computer Use" capabilities.
- Infrastructure Mastery: Proven ability to build and maintain distributed systems. You should be comfortable with container orchestration (Kubernetes) and cloud-native scaling.
- Telemetry & Observability: Experience building systems that record and replay complex user/system interactions (e.g., session recording, eBPF, or UI accessibility trees).
- Coding Excellence: Expert-level Python and strong proficiency in C++, Rust, or Go (essential for system-level components).
- Research Mindset: Ability to treat infrastructure as a research problem—running experiments on environment diversity, task difficulty, and data quality.
- ML Familiarity: Comfortable working with PyTorch and LLM inference frameworks to test how system-level changes impact model performance.
Why Join Us
- Work on one of the hottest problems in AI: Build and scale agentic AI systems that push the frontier of what autonomous, reasoning-driven software can do in the real world
- Own product and infrastructure end to end: Take ideas from zero → one → scale, shaping core product decisions, system architecture, and production infrastructure
- High-growth, high-impact team: Join a small, fast-moving team where your work ships quickly and materially impacts customers and the company's trajectory
- Direct collaboration with founders: Influence strategy and help define the long-term technical and product vision
- Startup upside and velocity: Experience rapid learning, ownership, and career acceleration
- Competitive compensation and benefits: Strong salary, 401(k), and comprehensive medical, vision, and dental coverage
- Flexible work culture: Hybrid work, unlimited PTO, and a focus on trust, autonomy, and outcomes
- Inclusive and collaborative environment: We value thoughtful debate, diverse perspectives, and building great things together—without ego
How to Apply
We encourage candidates who love tinkering with OS internals and are obsessed with the future of AI Agents to apply.
We are an equal opportunity employer and value diversity at our company.