// independent security research
Ievgen Bondarenko
I identify and validate vulnerabilities in AI model-serving, cloud, container, and open-source infrastructure.
01 / featured research
A Critical SSRF flaw let an unauthenticated caller cross the MLflow server network boundary through webhook redirects and DNS rebinding.
Unauthenticated input / Webhook delivery / Internal destination
SSRF, API-key exposure, and response forgery were validated in an AI SDK integration that accepted a caller-influenced provider endpoint.
Request endpoint override / Authenticated provider client / Attacker-controlled origin
LMDeploy enabled Hugging Face remote code by default during model loading, removing the operator decision that peer inference systems expose as an explicit opt-in.
Model reference / Inference initialization / Python remote code
02 / open source security
Container isolation
Isolation boundaries, networking, race conditions, checkpoint security, and protocol validation.
View evidenceModel-serving media retrieval
Security controls around multimodal media retrieval and shared connector policy.
View evidencePackage extraction
Archive extraction and symbolic-link boundary protection in package registries.
View evidenceCryptographic input validation
Strict JWK parsing and validation at a public cryptography-library boundary.
View evidenceProtocol parsing
Bluetooth protocol parsing and denial-of-service protections.
View evidenceDetection engineering
Detection content for suspicious network, identity, endpoint, and control-plane activity.
View evidence03 / research method
My research combines source-code analysis, data-flow tracing, controlled reproduction, exploit validation, impact analysis, and coordinated disclosure.
Target and trust-boundary analysis
traceSource review and data-flow tracing
traceReproducible test environment
traceProof-of-concept validation
traceImpact and exploitability assessment
traceVendor coordination
traceRemediation verification
verified04 / technical focus
05 / background
My security research is supported by hands-on experience in cloud security, detection engineering, incident response, enterprise infrastructure, and compliance-sensitive environments.
Read the background$ cat current_research
Currently researching trust boundaries and untrusted-input handling in AI-serving and evaluation infrastructure. Some findings are undergoing coordinated disclosure.
06 / security engagements
I review AI-serving systems, agent and evaluation infrastructure, and open-source components where untrusted input crosses a privileged boundary.
Remote model loading, resource retrieval, provider credentials, and the cloud or container boundaries around inference.
Tool trust, identity, untrusted inputs, provider endpoints, and the actions an evaluation harness can perform.
Source-led review of parsers, protocols, package handling, isolation controls, and dependency boundaries.
07 / contact
Open to AI infrastructure security roles, expert technical evaluation, open-source security engineering, and selective security research engagements.