Serguei Castillo serguei9090

Hi, I'm Serguei Castillo 👋

CTO & CIO Technology Leader | Enterprise Cloud Solutions Architect | SRE & DevOps Engineering Lead

⚡ Executive Summary & Leadership Impact

10+ years of technology and engineering leadership building, securing, and scaling high-availability enterprise platforms and multi-cloud architectures. Proven executive track record driving CTO product innovation and CIO enterprise governance: leading infrastructure operations, virtualization clusters, zero-trust security postures, FinOps cloud cost optimization, and telemetry-driven SRE practices.

🚨 Crisis Incident Command & Disaster Recovery (DR): Led emergency bare-metal recovery and fleet reconstruction during a catastrophic 95% enterprise outage/security incident—successfully restoring critical services within 24 hours and achieving full production operational recovery across all service tiers in 5 days.
💵 FinOps & Cloud Financial Governance: Implementing cloud unit economics, multi-cloud cost optimization (AWS, GCP, Azure), workload right-sizing, and cross-functional budget accountability between Engineering and Finance leadership.
🏢 Infrastructure & Platform Operations: Enterprise directory services (Active Directory AD DS, Group Policy GPO), storage clusters (TrueNAS ZFS, Ceph), fleet management (ManageEngine Endpoint Central, Windows/Linux), and hyperconverged virtualization (Proxmox, VMware ESXi).
🏛️ Architecture & System Blueprints: Authoring high-fidelity System Architecture Diagrams, Data Flow Diagrams, and Infrastructure Decision Records (ADRs) for stakeholders and engineering teams.
📊 Observability, SRE & Telemetry: Enterprise monitoring (Zabbix, Nagios, Prometheus, Grafana), deep network packet & flow analysis (Nmap, Wireshark) with PowerBI shareholder reporting, and 99.99% uptime delivery.
🔒 Security, Firewalls & Zero-Trust: Multi-tier perimeter security (Fortinet FortiGate, OPNsense, pfSense), Zero-Trust MFA/SSO (Authentik), full-stack enterprise mail server architecture (SMTP/IMAP with SPF, DKIM, DMARC enforcement), and SSL/TLS PKI.
🧠 Automation & Cloud AI Engineering: Enterprise AI orchestration (OpenAI APIs, Google Cloud AI / Vertex AI / Gemini Live, Ollama), full-stack automation (n8n, Python, PowerShell, Bash), and RESTful microservice architectures.

Note

🎓 Portfolio & Academic Research Statement

The projects published across this profile represent personal learning, open-source technical exploration, and academic research prototypes designed to explore advanced systems architecture, AI concepts, FinOps, and security engineering.

Status: Personal sandbox and technical portfolio showcase.
Educational & Research Focus: Engineered for personal skill development, academic exploration, and technical demonstrations.
Licensing: Individual repositories are published under their respective licenses—including MIT, GNU AGPLv3, and Source-Available terms—governing their community and production use.

🛠️ Technology Stack & Core Competencies

☁️ Cloud, Hypervisors & Storage Infrastructure

🏢 Enterprise Directory, Fleet Management & OS

🌐 Protocols, Routing, Mail Servers & High Availability

🔒 Firewalls, Reconnaissance, Cryptography & Identity

📊 Observability, Architecture Blueprints & Documentation

🤖 AI, API Architecture & Automation

🌟 Featured Engineering Projects

🚀 Cloud & AI Systems Architecture

The-Spatial-Eye

Real-Time Multimodal Spatial AI Platform & Infrastructure
Stack: TypeScript, Python FastAPI, Google Cloud Run, Terraform IaC, Firebase, WebSockets

Engineered an end-to-end cloud platform streaming bidirectional audio/video to Gemini Multimodal Live APIs over WebSockets.
Fully automated Infrastructure as Code (Terraform for GCP) with dual GitHub Actions CI/CD deploying edge builds and containerized Python backends to Cloud Run.
Enterprise zero-trust security architecture implementing client-side Bring-Your-Own-Key (BYOK) isolation.

code-rag

High-Performance Codebase Indexing & Semantic Retrieval Engine
Stack: Rust, Vector Embeddings, AST Parsing, CLI

High-throughput indexing and semantic search tool for large codebases.
Delivers sub-millisecond AST symbol extraction, vector distance ranking, and local token-efficient context retrieval.

🛡️ Enterprise Security & Cryptography

DualCrypt

Zero-Trust Multi-Party Threshold Encryption & Post-Quantum Escrow Platform
Stack: Rust (denc-core, denc-cli, denc-wasm), Tauri v2, React 19, Post-Quantum Crypto (ML-KEM / ML-DSA)

High-assurance zero-trust cryptographic platform eliminating single points of failure via $k$-of-$n$ Shamir Secret Sharing over $\text{GF}(256)$ and streaming AEAD (AES-256-GCM / XChaCha20).
Implements NIST FIPS 203 (ML-KEM-768 Kyber) and FIPS 204 post-quantum key encapsulation with air-gapped optical QR fountain mobile authentication and physical YubiKey hardware detection.
Cross-platform native releases (Windows, macOS, Linux, Android) with zero-knowledge browser WebAssembly client.

📊 Observability, SRE & Telemetry

InfraKitOps

Local-First SRE Workbench & Infrastructure Orchestration Cockpit
Stack: React 19, TypeScript, Tauri v2, Go 1.25, SQLite, Docker, Ansible, MCP

Engineered a local-first engineering workbench featuring 44+ offline utility modules, interactive Ansible playbook manager, and multi-step runbook automation engine with encrypted secret vaults.
Implemented strict hexagonal architecture (ports & adapters) coupling a zero-React pure TypeScript domain core to a Go desktop sidecar / containerized HTTP microservice.
Built-in network diagnostic suite (SNMP v1/v2c/v3, route mapping, packet flow inspection) and unified AI Hub with Model Context Protocol (MCP) tool execution gates.

LogLensAi & React-LogLensAI

High-Performance Production Log Analysis Cockpit
Stack: TypeScript, React, Telemetry Analytics, Regex Filtering

Production log inspection tool designed for SRE incident response, high-speed anomaly triage, and pattern recognition across distributed services.
Real-time client-side parsing capable of filtering large raw log streams without performance degradation.

🧠 Knowledge Graphs & AI Orchestration

GraphRAG-ollama-demo-chat

Knowledge-Graph Augmented RAG System for Local & Cloud LLMs
Stack: Python, Ollama, Graph RAG, Vector Search, Streamlit

Implemented structured knowledge-graph extraction coupled with vector search to drastically reduce LLM hallucinations in complex domain reasoning.

AuroraGraph & VibeGraph

Graph Intelligence & Relationship Visualization Systems
Stack: Python, NetworkX, Graph Topology, Data Analytics

Dynamic topological graph analyzers for visualizing multi-node connections, dependency trees, and cluster metrics.

🛠️ Developer Tooling & Enterprise Utilities

formfromfile

Self-Hosted Dynamic Schema Detection & Structural Form Engine
Stack: Go (Chi), React 19, TypeScript, SQLite / PostgreSQL, Firebase Auth, Docker

Full-stack self-hosted platform that ingests arbitrary XML/YAML/JSON/TOML schemas, automatically synthesizes validated interactive web forms, and exports bidirectional structured data with comment and key-order preservation.
Hardened with SSRF-guarded outbound webhooks, Cloudflare Turnstile bot protection, Argon2id authentication, and sub-second cold starts within a 19 MB distroless container.

mcp-websocket-transport

High-Performance Full-Duplex WebSocket Transport & Universal MCP Bridge
Stack: TypeScript, Python, Model Context Protocol (MCP), WebSockets, PyPI, npm

Universal full-duplex WebSocket transport layer for MCP, solving reverse sampling requests and progress stream timeouts over standard HTTP/SSE.
Dual-published libraries on PyPI and npm with 100% feature parity, zero-state persistent routing, and sub-millisecond framing overhead.
Includes mcp-ws-bridge CLI to seamlessly connect desktop LLM hosts (Claude Desktop, Cursor, Antigravity) to remote and containerized MCP servers.

AI-Bookmark-Manager

Local-First Intelligent Bookmark Manager & Chrome Extension with MCP Server
Stack: TypeScript, Chrome Extension (Manifest V3), Model Context Protocol (MCP), Chrome Native Sync, Local LLMs (Ollama / Gemini / OpenAI)

Intelligent, local-first Chrome Extension (Manifest V3) and workbench that automatically categorizes, tags, and summarizes bookmarks using local (Ollama, LM Studio) or cloud LLMs without third-party vendor lock-in.
Features Native Cross-Device Chrome Sync (.ai_bookmark_metadata shadow folder architecture) to synchronize AI-generated summaries, tags, and category blueprint prompts across machines at zero server cost.
Ships with a dedicated Model Context Protocol (MCP) Server (ai-bookmarks-mcp on npm), enabling AI coding assistants (Claude Desktop, Cursor, Antigravity) to query, search, and reference offline bookmarks directly from disk.
Built-in automated dead-link health scanner (404/500/timeouts), canonical URL deduplication, and non-destructive 10-state rolling snapshot undo history.