About Invyte, Inc

About Invyte.ai

Invyte.ai is an AI-native talent platform helping companies discover and hire high-caliber technical talent based on demonstrated capability, not just résumés. Candidates are evaluated through expert AI interviews, practical assessments, and evidence of how they solve real problems.

About the Opportunity

Invyte.ai is working with an early-stage Y Combinator-backed AI company to hire a Founding Engineer — Agent Evals & Environments.

The company is building autonomous agents for complex enterprise operations, with a strong focus on IT service management and enterprise systems. The team is tackling the difficult infrastructure around agent evaluation, realistic environments, benchmarking, and reliable execution of multi-step enterprise workflows.

This is an opportunity to join a small founding team and work directly on the systems required to determine whether AI agents can reliably perform consequential real-world work.

Job Description

We're looking for a Founding Engineer to build evaluation and simulation infrastructure for autonomous AI agents operating across complex enterprise workflows. A core question you'll help answer: Did the agent actually complete the task correctly, reliably, and safely? You'll work closely with the founding team on agent environments, evaluation systems, benchmarks, and infrastructure for understanding where frontier models succeed and fail. This is not primarily a prompt-engineering role. You'll build the environments and measurement infrastructure needed to determine whether autonomous agents can reliably perform consequential real-world work.

Key Responsibilities

Build realistic, stateful environments for evaluating autonomous agents.

Turn real-world workflows and agent failures into reproducible evaluation tasks.

Design deterministic graders, verifiers, datasets, and gold-standard trajectories.

Run evaluations across models, agent architectures, and harnesses.

Analyze agent traces, tool calls, and trajectories to identify failure modes.

Build infrastructure for repeatable evals, regression testing, and benchmark analysis.

Design evaluations resistant to reward hacking and benchmark gaming.

Translate messy enterprise workflows into measurable agent tasks.

Qualifications

Have built and shipped LLM agents or multi-step tool-using systems.

Have experience with agent evals, benchmarks, graders, simulations, or RL environments.

Are strong in Python and backend/infrastructure engineering.

Understand tool calling, planning, state management, context management, and agent orchestration.

Enjoy investigating why agents fail, not just getting demos to work.

Care deeply about rigorous measurement and reproducibility.

Experience with complex enterprise workflows or systems such as ITSM, developer tooling, security, support, or operations is a plus.