Infrastructure & software engineerBuild. Deploy.
Build. Deploy.
Keep it running.
I’m Andy. I build software and operate the systems behind it—from blockchain validators and cloud infrastructure to tools for AI agents.
visitor@andy: ~/career
Welcome to my corner of the internet.
Type help to look around.
whoami
Andy Sustic
Infrastructure & software engineer. Eight years in production systems.
Blockchain networks. Cloud platforms. Connected devices.
Now building AgentVouch and exploring local AI infrastructure.
↑ ↓ historyTab autocompleteEnter to run
A few things
behind the work
behind the work
Internet Computer Node onboardingParaFi Technologies
10,000
Concurrent device connections
GTT · Transit infrastructure01 / Career archive
A closer look at the work.
The roles, projects, and details
behind the one-page introduction.
6 roles · 5 projects
Experience
2018 — 2026 · Earlier experience includedRemote
- Led Solana validator deployment and scaling from testnet to mainnet on dedicated bare-metal servers. Maintained uninterrupted operations from Q2 2025 through the role, with approximately 184,000 SOL in delegated stake at peak.
- Secured acceptance into the Solana Foundation Delegation Program and independent staking programs through sustained validator performance and operational reliability.
- Designed and implemented active-passive validator failover on Linux with Bash and Ansible. Automated checks for vote lag, missed slots, and node health enabled rapid recovery to a hot standby.
- Integrated Jito’s BAM client and operated infrastructure as a node operator in the Jito TipRouter Node Consensus Network.
- Deployed and operated a standalone public Solana RPC node on Latitude bare metal, using Linux, Bash, and Ansible for management and Grafana and Prometheus for monitoring.
- Automated validator and RPC deployments with Ansible. Managed Kubernetes workloads on GCP and Terraform-managed GKE, routing infrastructure and workload security logs through Fluent Bit into Google Cloud Logging.
- Built Grafana dashboards and Prometheus monitoring and alerts for vote credits, validator performance, cluster health, and infrastructure availability.
- Led ParaFi’s end-to-end onboarding as an ICP node operator, coordinating technical, operational, legal, and documentation requirements.
Tools & technologies
LinuxAnsibleTerraformGCP / GKEKubernetesDockerPythonBashGoTypeScriptSQLPrometheusGrafanaFluent BitSolanaEthereumICPJito
Remote
- Supported on-premises and AWS EC2 deployments with Ansible and custom CLIs, later adopting Pulumi for infrastructure automation.
- Investigated and resolved infrastructure, deployment, and application issues using Grafana, Elastic APM, Splunk, and New Relic.
- Ran existing GitHub Actions workflows, coordinated releases through Slack, and performed manual database releases on Windows.
- Developed TypeScript and Node.js tooling to refresh infrastructure inventory used by observability platforms.
- Drove New Relic adoption across the SRE team and helped standardize monitoring workflows.
- Participated in production on-call support and assisted engineering teams during customer-impacting incidents.
Tools & technologies
AnsiblePulumiAWS EC2GitHub ActionsWindowsTypeScriptNode.jsGrafanaElastic APMSplunkNew Relic
Remote
- Developed Python ETL pipelines for internal operational data, blockchain transactions, staking rewards, and validator-performance datasets.
- Built a self-hosted KPI dashboard proof of concept with Metabase and PostgreSQL in Docker, giving engineering teams access to operational and business metrics.
- Worked with Pandas, Beautiful Soup, APIs, AWS Lambda/S3, and web tooling to collect and transform fragmented data into usable reporting.
Tools & technologies
PythonPandasBeautiful SoupPostgreSQLMetabaseDockerAWS Lambda / S3
Boulder, CO · Now S&P Global
- Developed and maintained a C# internal platform for managing teams and executing DevOps operations, including server-maintenance workflows using JavaScript, AJAX, Bash, and Rundeck.
- Extended existing Chef recipes and authored new recipes and health checks across Windows and Linux to identify configuration-management failures.
- Led the migration of web-application monitoring to a Prometheus, ELK, and Grafana stack, improving observability across supported applications.
- Worked in Agile Scrum teams through planning, backlog grooming, demos, retrospectives, and production support.
Tools & technologies
C#JavaScriptAJAXBashChefRundeckWindowsLinuxPrometheusELKGrafana
Oakdale, MN
- Upgraded the production MQTT platform to VerneMQ, supporting 10,000 concurrent bus-device connections transmitting location and timestamp telemetry, including in San Francisco.
- Used Docker, HAProxy, and TLS/SSL termination in the messaging infrastructure.
- Built Python machine-learning, data-processing, and visualization tools to analyze Transit Signal Priority effectiveness using traffic data hosted in AWS.
- Processed multi-gigabyte datasets into machine-learning-ready data and customer reports for SFMTA.
- Built Python AWS Lambda functions to ingest and transform traffic data, working with S3 and DynamoDB.
- Collaborated with developers and quality engineers in two-week sprints on transportation and emergency-response use cases.
Tools & technologies
PythonPandasMQTTVerneMQDockerHAProxyTLSAWS LambdaS3DynamoDB
Minnesota
- Woodbury Athletic Association: organized registration and management for thousands of student athletes across dozens of sports.
- Shane Co., Men’s Wearhouse, Target, and Domino’s: customer service, sales, inventory, reporting, conflict resolution, and fast-paced operations.
Tools & technologies
Customer supportOperationsRegistrationInventoryReporting
Independent projects
Built from curiosityIndependent · Open sourceView source
- Designed and built AgentVouch from scratch, deploying the public Next.js application to Vercel and the protocol to Solana devnet.
- Implemented a 25-instruction Solana Anchor program in Rust covering agent registration, USDC-backed vouching and author bonds, skill listings and purchases, x402 settlement, dispute resolution, voucher slashing, refunds, and revenue distribution.
- Built the TypeScript and Next.js marketplace, agent-facing APIs, and CLI with wallet authentication, on-chain transactions, x402-gated downloads, Neon PostgreSQL persistence, versioned skill publishing, AI-assisted security scanning, and entitlement controls.
- Developed and tested a gated Solidity and Foundry implementation for Base Sepolia.
- Built GitHub Actions CI, Vitest, Anchor and Foundry test suites, generated clients, deployment tooling, database migrations, smoke tests, and operational runbooks.
Tools & technologies
TypeScriptNext.jsRustAnchorSolana / SVMSolidityFoundryPostgreSQLVercelx402
Independent research
- Built and benchmarked local inference environments on a MacBook Pro with M1 Pro and an NVIDIA RTX 3090, using MLX, Ollama, llama.cpp, and LM Studio.
- Exposed gpt-oss, Gemma 26B A4B, and Qwen 28B models through local endpoints.
- Compared quantization, memory usage, performance, and deployment tradeoffs with LM Studio tooling and AI-assisted test workflows.
Tools & technologies
MLXOllamallama.cppLM StudioApple siliconNVIDIA RTX 3090Open-weight models
Independent · Open sourceView source
- Founded and maintained a privacy-focused blockchain network, managing the codebase, Linux infrastructure, community operations, and contributor workflows.
- Built and operated seed nodes, a block explorer, mining-pool infrastructure, Grafana monitoring, and ecosystem tooling using C++, Python, Node.js, Docker, and Bash.
- Managed releases, technical support, and community communications across GitHub, Discord, and public forums.
Tools & technologies
C++PythonNode.jsBashLinuxDockerGrafanaGitHub
Open sourceView source
- Built a Python trading bot around the Coinbase API.
- Maintained open-source infrastructure for the WHACKD Foundation, alongside smaller Web3 and automation experiments.
Tools & technologies
PythonCoinbase APIAutomationNetwork infrastructure
Team projects
- Placed 4th of 22 teams in the Minneanalytics 2018 data-science competition.
- Created real-time manufacturing telemetry in a 24-hour hackathon.
Tools & technologies
Data scienceTelemetryPrototyping
02 / Capabilities
The tools behind the systems.
From production operations
to independent experiments.
Infrastructure & delivery
Deploy, automate, and operate.
LinuxBare metalTerraformAnsiblePulumiDockerKubernetes / GKEAWSGCPGitHub ActionsCI/CDHigh availability
Observability & data
Make system behavior visible.
GrafanaPrometheusFluent BitCloud LoggingNew RelicSplunkElastic APMELKPostgreSQLMetabasePandas
Languages & software
From automation to application code.
PythonBashGoTypeScriptJavaScriptC#C / C++RustSoliditySQLNode.jsNext.js
Protocols & networks
Stateful, distributed workloads.
Solana validators & RPCEthereumICPJito BAMJito TipRouterAnchorMQTTVerneMQHAProxyTLS
AI tools & local inference
Development workflows and independent lab work.
CodexClaudeCursorHermesMLXOllamallama.cppLM StudioOpen-weight model deployment
03 / The blog
Field notes.
Notes on infrastructure, software, and lessons from the work.
Should You Let Your Agent Install That Skill?
Trusted agent skills, reputation, economics, and security.
6 min readRead note
The Skills Supply Chain Attack Nobody Is Talking About
How AI agent skills create supply-chain risk, and how USDC-backed author reputation, file review, and sandboxing help inform safer installation decisions.
7 min readRead note
Junk Skills: When SKILL.md Gets Ahead of Reality
Why AI skill descriptions need evidence: inspect author reputation, USDC-backed vouches, and dispute history before trusting a SKILL.md file.
3 min readRead note
Education
University of Wisconsin–River Falls
B.S. Computer Science & Information Systems
River Falls, Wisconsin