Software • Cybersecurity • AI Developer — network infrastructure, security tooling, and LLM-powered automation
- AI + Cybersecurity: LLM API integration, multi-stage analysis pipelines, security review assistants, automated reporting & remediation
- Security Engineering: recon + API discovery, pentesting-oriented tooling (educational), exploit concepts, detection-minded engineering
- Network Infrastructure: reverse proxies, client-server systems, production networking services
- Developer Tooling: CLI/GUI utilities, workflow automation, version-control experiments
| Project | What it does | Stack | Link |
|---|---|---|---|
| SentinAI (Security Review Assistant) | AI-assisted secure code review: heuristic hotspot detection + cross-file context indexing + DeepWiki-style project semantics, with multi-layer LLM (Qwen → DeepSeek) to converge on actionable findings + fixes (+ optional PoC generation) | Python, Flask, FastAPI, LLM APIs | not public yet |
| Simple-Ai-agent | Lightweight AI agent scaffold focused on LLM API connectivity and tool-ready automation workflows (a base for assistants, pipelines, and security-oriented agents) | LLM APIs, Automation | https://github.com/v7ren/Simple-Ai-agent |
| Logeral (Clerk-style Auth Platform) | Full auth provider: Next.js dashboard + Flask core auth backend. Create apps, issue pk_/sk_ keys, support BYOK or Hosted OAuth, sessions, scoped apps, and an SDK-style client integration | Next.js, TypeScript, Flask, SQLite, OAuth | not public yet |
| Certion-ReverseProxy | Production-grade reverse proxy for the Certion platform with modern architecture | TypeScript, Node.js, Networking | https://github.com/v7ren/Certion-ReverseProxy |
| api-endpoint-discoverer | Multi-phase API endpoint discovery & detection for security testing and research | Python, API Analysis | https://github.com/v7ren/api-endpoint-discoverer |
| vmc2 | Modular reverse shell framework for security research and educational purposes | Python, Networking | https://github.com/v7ren/vmc2 |
| verctrl | Lightweight version control CLI + PyQt6 GUI | Python, PyQt6 | https://github.com/v7ren/verctrl |
| Domain | Technologies | Level | Notes |
|---|---|---|---|
| LLM Engineering | LLM APIs, streaming, structured JSON outputs, cost/token control, multi-stage pipelines | Experienced | Security-focused agents + automation |
| AI + Security | hybrid analysis (heuristics + indexing + LLM), report generation, remediation guidance, optional PoC workflows | Experienced | SentinAI |
| Network Infrastructure | TypeScript, reverse proxies, client-server architecture | Advanced | Certion-ReverseProxy |
| Security Research | Python, reverse shells, API discovery, pentesting workflow | Intermediate | vmc2, api-endpoint-discoverer |
| Developer Tooling | Python, CLI, PyQt6 GUI, Git/workflows | Intermediate | verctrl |
| Backend Development | Python/TypeScript, API design, service architecture | Intermediate | auth + security services |
SentinAI is an AI-assisted secure code review system designed to behave less like “rule spam” and more like a security engineer:
- Understands the project first (components, entrypoints, dependencies, data flow)
- Finds high-risk hotspots cheaply (keywords/regex/symbol indexing)
- Uses LLMs for context-aware verification (reduce false positives, add CWE/confidence/impact/remediation)
- Optional PoC generation for validating findings in a safe test environment
Architecture (Gateway + Worker):
main4.py— Flask API gateway/orchestrator (uploads, scan_id, SSE streaming, history DB, docs/PoC endpoints)nodev17.py— FastAPI worker (tiered scans, symbol/call graph lite index, DeepWiki build, PoC enrich)LLMindexer.py— DeepWiki-like semantic indexer (project IR + multi-chapter markdown)
Tiered scanning model (cheap → expensive):
- Tier 1 (Qwen/DeepInfra): high-recall prescan
- Tier 2 (DeepSeek): context validation + CWE/confidence/impact
- Tier 3 (DeepSeek + streaming): consultant-grade report (JSON repair for stream truncation)
- Tier 4 (optional): PoC generator bound to stable
fingerprint
Logeral-server: Next.js platform UI + Flask auth backend. Create applications, issue pk_/sk_ keys, support BYOK or Hosted OAuth, sessions, and per-app user DB scoping.
Logeral client/SDK: Next.js client + logeral-auth package (hooks, components, server utils, middleware) for protecting routes with cookie-based sessions.
- Server: not public yet
- Client/SDK: not public yet
All security-related projects are for educational and research purposes only. I do not support illegal or unauthorized activity.