# Giulio Micheloni - Senior Infrastructure Engineer

**Role**: Senior Infrastructure Engineer
**Contact**: giulio.micheloni@gmail.com | [LinkedIn](https://linkedin.com/in/giuliomicheloni) | [GitHub](https://github.com/gmichelo)

## Summary
Senior infrastructure engineer specializing in multi-cloud and on-prem Kubernetes clusters, distributed backend systems, and advanced observability. These days architecting scalable and secure agentic AI platforms.

---

## Core Expertise
- **Cloud & Infrastructure**: AWS (EKS, IAM, MSK, RDS, Bedrock), GCP (GKE, Cloud SQL, Workload Identity), Azure (AKS, ARM), Firecracker microVMs
- **Kubernetes & Platform**: EKS, AKS, GKE, Flux CD, Karpenter, KEDA, Contour / Envoy, Kyverno
- **Observability**: OpenTelemetry, HyperDX, Prometheus, Grafana, CloudWatch, Loki, eBPF
- **CI/CD & Developer Tooling**: GitHub Actions, Buildkite, sccache, BuildKit, Terraform, Helm
- **Databases**: PostgreSQL, ClickHouse, Memgraph, Kafka
- **Security & Networking**: IAM / RBAC, Cognito Federation, PrivateLink, TLS / rustls, Container Hardening, SOC2 Compliance
- **Languages**: Rust, Go, Python, TypeScript / React, Bash

---

## Professional Experience

### Senior Infrastructure Engineer - Ciroos.ai
*(August 2025 - Present)*
- Architecting scalable and secure Agentic AI platforms.
- Built a Rust ETL pipeline on Clickhouse to build infrastructure's topology graph and feed the Ciroos.ai LLM knowledge graph.
- Owner the internal infrastructure automation with Terraform and GitHub Actions.
- Instrumented Ciroos.ai backends with OpenTelemetry and HyperDX to provide distributed tracing and observability.
- Owned the internal multi-arch build orchestration system, accelerating Rust CI builds from >30m to <10m with sccache and concurrent BuildKit execution.
- Developed AWS, GCP and Azure MCP servers, introducing streamable HTTP transport to enable horizontal autoscaling for MCP servers.
- Hardened cloud security by mitigating Envoy/Python CVEs, securing database connection layers, and implementing granular RBAC for cluster access.
- Implemented KEDA-based proxy autoscaling utilizing Prometheus metrics and configured Karpenter NodePools to optimize cluster resource utilization.

### Senior Infrastructure Engineer - Namespace Labs
*(January 2023 - July 2025)*
- Engineered large-scale CI/CD orchestration systems leveraging Firecracker microVMs to deliver high-performance, isolated build environments.
- Optimized multi-tenant infrastructure and distributed orchestration components, significantly reducing build times for containerized workloads.
- Developed core platform features focused on cloud automation, developer productivity, and advanced backend systems integration.

### Senior Infrastructure Engineer - Amazon Web Services
*(August 2021 - January 2023)*
- Built and maintained critical eBPF-based TCP monitoring tools and performance regression pipelines to ensure high network reliability.
- Designed scalable cloud-native infrastructure solutions, focusing on internal platform observability and network telemetry data ingestion.
- Strengthened system resilience through comprehensive automated testing and integration with native AWS monitoring ecosystems.

### Senior Infrastructure Engineer - Cisco
*(January 2018 - July 2021)*
- Developed hybrid-cloud Software-Defined Networking (SDN) controllers, bridging on-premise infrastructure with public cloud environments.
- Directed technical implementations alongside distributed contractor teams to deliver highly available networking products.
- Implemented core distributed systems functionality, ensuring robust failover and state synchronization.

---

## Technical Writing & Thoughts
I occasionally write technical articles detailing my operational experiences building complex systems:
- **Designing reliable CI/CD and ephemeral runners at scale**: Architecting ephemeral environments across thousands of builds utilizing Firecracker microVMs and smart workload scheduling.
- **Practical lessons from Kubernetes, observability, and AI infrastructure**: When merging AI workloads with Kubernetes, typical multi-tenant best practices often break down. Here's how to scale GPU deployments efficiently.

---

## Selected Flagship Projects

### SanoMacro.com
A full-stack health application in production. I built a multi-stage LLM RAG pipeline to classify food text queries into precise, verifiable macronutrient data vectors.
- **Stack**: React, TypeScript, Node, Redis

### Personal AI GPU Lab
Managed a bare-metal custom Kubernetes GPU cluster running distributed vLLM inference nodes for hosting my own private local AI coding assistants, highly monitored with Prometheus and Grafana.
- **Stack**: Kubernetes, vLLM, Grafana, Bare-metal GPU

### K8s Network Prober
An open-source custom Kubernetes Operator built from scratch in Go for natively exporting precise RTT network telemetry metrics continuously via Prometheus integration.
- **Stack**: Go, Kubernetes Operator, Prometheus