I build infrastructure at the intersection of AI and security. My current focus is on securing AI agent systems — specifically MCP (Model Context Protocol) servers that connect LLMs to real-world tools.
Previously: data engineering pipelines with Apache Spark, ML systems for traffic prediction, legal NLP, and clinical AI.
Real-world security fixes credited: CVE-2026-14540 in Google's official MCP Toolbox, and a merged hardening fix in GitHub's own official MCP server.
Security scanner for MCP servers — the Snyk for AI agent tooling
Detects prompt injection in tool descriptions, credential leaks (25 patterns), exposed endpoints, and tool poisoning before you connect MCP servers to your AI agent. 66 detection rules, CVSS-scored findings, HTML reports. Credited with CVE-2026-14540 in Google's official MCP Toolbox.
pip install mcp-safeguard🌐 syedanas01.github.io/mcp-safeguard
Replication package: does reporter attribution survive into CVE records, GitHub's OSV export, and the NVD? Four datasets, captured and documented.
Multi-scope AI content analysis network — NLP pipeline for multi-domain classification.
Real-time taxi demand forecasting using Apache Spark MLlib — big data streaming pipeline for hotspot prediction.
End-to-end ML pipeline for flight delay prediction and root-cause analysis.
CNN + XGBoost hybrid model for early Parkinson's detection from clinical biomarker data.
| Domain | Tools |
|---|---|
| AI / LLM | Claude API, FastMCP, LangChain, OpenAI |
| ML / Data Science | PyTorch, scikit-learn, XGBoost, pandas |
| Big Data | Apache Spark, Hive, AWS EMR |
| Security | CVSS scoring, OWASP, prompt injection detection |
| Observability | Prometheus, OpenTelemetry, Grafana |
| Languages | Python, TypeScript, SQL, Shell |
| Infra | Docker, GitHub Actions, AWS |
- 📧 anasmohiuddinsyed@gmail.com
- 💼 Open to AI/ML engineering and security research roles
- 🌐 github.com/SyedAnas01