Dublin, Ireland · UTC+0/+1 · remote-ready
I'm Jose Elias Costa, an IT support engineer with over a decade across Microsoft 365, device fleets, networks and data centre operations, including hardware lifecycle work at AWS Dublin.
Alongside that, I design and run production Python services with on-chain payment rails: APIs that software agents buy from by the request, settling on three networks. Both halves are on this page, because most of the useful work sits where they meet.
Three bands of work. The first two are what a decade of practice bought; the third is where I spend my own time.
The everyday layer that a growing company outgrows before it hires for it.
Microsoft 365, Entra ID and Google Workspace: sane defaults, MFA, conditional access, licence hygiene, and shared drives that don't turn into a maze.
Provisioning at scale across macOS, Windows and ChromeOS. Intune policies, asset lifecycle, and a record of what exists and who holds it.
Repeatable joiner/leaver flows covering accounts, devices and access, so a new hire is productive on day one and a departure closes cleanly the same day.
The named person for a 5–50 person office: Tier 1–2 support, vendor coordination, ticket triage in Jira, and SLAs actually measured.
Runbooks, standard operating procedures and internal guides written so a non-technical colleague can follow them without calling anyone.
Native Portuguese and fluent English, useful for remote-first companies with Brazilian or wider LATAM user bases, or teams expanding into the EU.
Where support work stops being manual. Scripts, monitoring and the plumbing that decides whether a service is reliable or merely lucky.
Python, PowerShell and Bash applied to the tasks that quietly eat a week each month: provisioning, reporting, patching, reconciliation.
Watchdogs that page a human for real failures and stay silent otherwise, including the tuning it takes to stop an alert being ignored.
Encrypted, off-site, and restore-tested. That last part is the one usually skipped, and it is what proves the backup works before anyone needs it.
systemd, nginx, Cloudflare (DNS, WAF, Tunnel), TLS, and small VPS setups that survive a reboot without anyone remembering the steps.
Collectors that pull and normalise data continuously, built so a rate limit or a slow upstream degrades the feed instead of taking the service down.
A focused read of what you already run: failure modes, security surface, what your metrics are really saying, plus a prioritised list of what to fix first.
Certified in Web3, active on-chain daily with my own capital, and running a production API that software agents pay for by the request.
HTTP 402 end to end: price manifests, signed payment envelopes, facilitator settlement, replay protection and per-network accounting.
USDC across Base, Solana and Algorand, including the awkward parts, where one network's payload format quietly breaks another's clients.
Building for a customer that is a program: discovery manifests, machine-readable docs, MCP servers, and pricing a purchasing agent can parse.
Front-line support for exchanges and wallets from someone who trades daily: failed transactions, bridging, self-custody, funding rates, API connectivity.
Inner City Scooters, Dublin
Tier 1–2 support end to end for onsite and remote teams, exceeding SLA targets by 20%. Device fleet, account provisioning and asset lifecycle across warehouse operations; streamlined joiner/leaver workflows that cut setup time and access-related tickets.
Amazon Web Services, Dublin
Hardware lifecycle for large-scale cloud infrastructure: installation, diagnostics and decommissioning of network equipment under strict change procedures, supporting 100% uptime targets.
Inner City Scooters, Dublin
First tenure in the same role: support, onboarding processes and hardware configuration. Rehired in 2025.
LDC Soluções & Indra, Brazil
1st and 2nd line support for software and network infrastructure at an international IT consultancy and a regional provider. 95% of tickets resolved within target; authored the SOPs the team worked from.
Built and operated solo, running in production today.
An on-chain data API sold per request, in USDC, over the x402 protocol. No accounts, no email, no KYC. The customer is a wallet address. Everything is mine end to end: data collection, pricing manifest, payment verification, settlement across three networks, public metrics, and the alerting that wakes me when the money path breaks.
A gateway that lets an existing API sell by the request without touching its code. Non-custodial by design, with funds never passing through it, so it adds metering and settlement without becoming a money service.
Paper-trading agents on the same infrastructure, sharing one exchange rate limit across three independent processes through a budgeting client. An exercise in making autonomous systems share a scarce resource without starving each other.
Whether it's a Microsoft 365 tenant nobody owns, a manual process eating a week a month, or an API you'd like software agents to be able to pay for, send a note describing the situation. I read everything and reply to anything concrete.