Research Engineer (AI Pentest)
Job Description
馃實 We鈥檙e making security suck less for developers.
Security tools haven鈥檛 kept up with how software is built today. They interrupt teams, slow releases, and turn security into a bottleneck instead of a capability. We built Aikido to change that. Aikido builds developer-first security products that reduce real risk without getting in the way of shipping software. We focus on what actually matters and automate the rest.
We鈥檙e taking on legacy security tools teams have been stuck with, and we鈥檙e winning. If you want to help us take market share and build products developers actually enjoy using, you鈥檙e in the right place.
Founded in 2022 by third-time founders, Aikido has $85M in the bank and a long runway ahead. We鈥檙e building toward self-securing software. Join an all-star team. Take real ownership. Push boundaries. Build things that matter.
We鈥檙e looking for a Research Engineer to join our AI Pentest team and build the feedback loops that make our pentesting agents improve over time. You鈥檒l use traces, evals, experiments, and production data to incrementally improve the agent harness and architecture, turning agent failures into measurable gains in vulnerability discovery, verification, and reporting.
Responsibilities
Make our agents measurably better at finding, verifying, and reporting vulnerabilities.
Design automated feedback loops that convert agent traces and failures into improvements of the agent harness.
Run data-driven experiments with different agentic architectures, workflows, tools, prompts, and eval strategies.
Build evals, benchmarks, and regression tests to measure progress and prevent regressions.
Improve the reliability, performance, accuracy, and cost-efficiency of our AI pentesting platform.
Ship validated improvements to production and own them end-to-end.