I build secure AI systems — gateways, policies, audit trails, and detection content that make LLM traffic safer and more visible.
Currently focused on:
- 🔐 AI security gateways — policy enforcement, secret scanning, agent tool guardrails
- 🧠 LLM systems — routing, RAG, privacy-preserving audit logging
- 📡 Detection engineering — SIEM rules, Sigma, dashboards for AI workloads
- ⚙️ Backend & APIs — FastAPI, Docker, REST services
- 🌐 Full-stack — React, TypeScript (Refine, MUI)
A self-hosted AI security gateway for developers and teams.
AIWall sits between your apps and AI providers. You get visibility, policy enforcement, secret scanning, audit logging, cost tracking, and agent tool guardrails — without routing traffic through someone else's cloud.
What ships today
- OpenAI-compatible proxy for clients, scripts, and coding tools
- Secret scanning before prompts hit a provider
- YAML policies (allow / warn / block / redact) with a control panel
- Agent tool guardrails (risky shell/file actions → approve/deny)
- Privacy-preserving audit log + cost tracking
- Alerts via Telegram, webhook, or ntfy
Sister repos
| Repo | Role |
|---|---|
| AIWall | Core gateway |
| AIWall-detections | Wazuh / Sigma / Grafana packs |
| AIWall-redteam | Adversarial tests and mitigation checks |
Reference architecture for securing clinical AI end-to-end: attack → gateway → audit → SIEM detection → dashboard, using only synthetic healthcare data.
Covers PHI-aware filtering, structured audit logging, MITRE ATLAS–mapped detections, and a red-team program with remediated findings.
Backend & security
- Python, FastAPI
- Policy engines, audit logging, REST APIs
- Docker
AI / LLM
- LLM gateways and routing
- Embeddings, vector search
- Privacy / PHI-aware filtering
- PyTorch, NLP
Detection & observability
- Wazuh, Sigma
- Grafana / Loki-style log pipelines
Frontend
- React, TypeScript
- MUI, Refine
- Vite
Data
- SQL, Pandas, NumPy
Tooling
- Linux, Git, Bash
- AIWall — self-hosted AI security gateway
- MedSecLab — clinical AI security reference architecture
- AIWall-detections — detection packs for AIWall audit events
- AIWall-redteam — red-team payloads and regression checks
Older repositories reflect earlier work in data science and analytics.
Email: Mohsen.Bahremani@proton.me
