← All speakers

Bio, Work & Ideas

Sarah Khalife

Conference affiliation: GitHub · 2024

Sarah Khalife is an application-security and developer-platform specialist who helps enterprise engineering teams find and fix vulnerabilities within their existing software-development workflows. Her work spans cloud-native deployment, financial-services security, and AI-assisted vulnerability remediation, with a particular focus on making security alerts actionable for developers.

Khalife began as a senior software engineer at GE, working on cloud applications and platforms, before joining GitHub as a solutions engineer. Her early GitHub work focused on automating Kubernetes application testing with kind and GitHub Actions, creating consistent test environments for individual pull requests. She also wrote about adopting containerized deployments without disrupting existing tools, team practices, or delivery processes.

By 2022, she was applying that workflow-first approach to financial services, combining security testing, builds, and deployment within a single pull request. She represented GitHub at a Fintech Open Source Foundation governing-board meeting in January 2024 and was a principal solutions engineer serving financial-services customers when she appeared at AI Engineer World’s Fair later that year.

Making security findings useful

Khalife’s approach to AI-powered application security emphasizes four practical priorities:

  • Shared security ownership. Security checks belong in repositories and pull requests, giving developers and application-security teams common visibility before release deadlines turn unresolved findings into emergencies.
  • CodeQL-guided vulnerability remediation. CodeQL data-flow analysis traces vulnerable inputs through multiple paths to a shared endpoint, supplying context for suggested fixes. For cross-site scripting, an autofix might recommend an HTML-escaping dependency; developers review the change and rerun scans before accepting it.
  • AI-assisted secret detection. Generated custom patterns and contextual analysis can identify exposed credentials and unstructured passwords, while push protection blocks secrets before they enter repository history. Previously exposed credentials require revocation or rotation, and broader scanning must account for false positives.
  • Repository-wide security coverage. Alert counts alone cannot reveal which code is unprotected. Khalife highlights repositories that never pass through conventional CI pipelines and therefore escape scanning, making organization-wide visibility essential.

Read the topics behind these talks

1 conference talk

References