SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Wispr AI is building the voice interface for computing. The company's products—Flow (natural voice interaction in any application) and Notetaker (context-building across conversations)—represent a shift toward AI agents that can perceive, understand, and act with earned trust.
In this role, you will be the first engineer to build the verification and evaluation layer that allows Wispr's AI agents to determine when their work is complete with confidence. Currently, agents write code and respond to incidents autonomously, but every loop ends with human verification. You'll eliminate that bottleneck by creating systems that let agents self-verify.
Key responsibilities:
- Design and build end-to-end verification loops that enable agents to confidently determine task completion
- Create fast, reliable testing environments and convert failures and human feedback into durable evaluation signals
- Improve how context, instructions, skills, and memory are delivered to agents, and measure what drives performance gains
- Operate the agent fleet as a production system, measuring quality, adoption, and real-world impact
You'll work hands-on across application code, infrastructure, and unfamiliar systems. This is an ambiguous, high-ownership problem that demands empirical rigor and attention to subtle failure modes.
The team is talent-dense, holds strong opinions, tests them quickly, and prioritizes the final human experience alongside technical excellence. Prior experience in voice, model training, or Wispr's product domains is not required.
Qualifications:
- Built internal tools that other engineers adopted and can articulate how you measured their impact
- Deep, hands-on experience with agentic coding tools and specific opinions about their strengths and limitations
- Experience building evaluation or verification infrastructure (test harnesses, CI systems, eval pipelines, benchmarks)
- Ability to work hands-on across application code, infrastructure, and unfamiliar systems
- Empirical mindset: attentive to subtle failure modes, comfortable owning ambiguous problems
Visa sponsorship available (H1B, O1, EB1, L1, STEM OPT, and others).
