Skip to content
View roselis-lab's full-sized avatar
😍
Working (a lot)
😍
Working (a lot)

Block or report roselis-lab

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
roselis-lab/README.md

Hi, I'm Aleksander

AI Security MLSecOps MCP Threat Modeling

Product security engineer working on AI / LLM systems. My focus is the architectural layer: MCP, agentic platforms, threat models, and the tooling around them.

Current focus

  • Threat modeling for LLM, agentic, and MCP systems
  • MCP security and the design of agentic platforms
  • Automated LLM application testing (Promptfoo, prompt injection, jailbreak, data leakage)
  • AI governance and SSDLC for GenAI in regulated environments

Selected work

  • Predixor — open source. Generates security requirements from a machine-readable description of an AI architecture.
  • MCP gateway & MCP platform security — architecture and requirements for safe MCP deployment in an enterprise, done with a platform working group. Defended at architectural review.
  • LLM / Agent / MCP threat models — co-authored, with mitigations mapped to specific components and owning teams. The MCP slice was built from primary research.
  • Promptfoo integration for LLM security testing — designed and shipped automated security testing for internal LLM apps; integrated with platform and DevSecOps pipelines.
  • Vibecoding & Shadow AI — policy and controls for AI-assisted development. Co-authored a mandatory AI-security course for engineering.

Background

Currently AI / MLSecOps in a large bank. Before that I launched and ran the InfoSec practice at a B2B services company (express audit, OWASP, incident response). Earlier I did infrastructure security work: NGFW, SIEM, EDR, audit automation in Python / Bash / PowerShell. Bachelor in Information Security (SUAI), pentest training, English C1/C2.

Guest expert at ITMO Talent Hub and reviewer on the ITMO master's thesis defense commission. Talks at ProductStar Online School.

Contact

Telegram Email

Pinned Loading

  1. predixor predixor Public

    Deterministic threat modeling engine for GenAI architectures. Predicate-based library maps applicable threats and required controls to your product graph

    TypeScript 4

  2. POC-Banking-ML-Attacks POC-Banking-ML-Attacks Public

    Minimal reproducible PoC of 3 ML attacks (adversarial, extraction, membership inference) on a credit scoring model. Includes pipeline, visualizations, and defenses

    Python 3