I build with AI, end to end.

I’m Stratos Theodorou, an AI-enabled, product-minded engineer and technical co-founder of WeatherXM and Ex Machina, in Athens. I work with AI on everything: a company agent that ships features, AI that customers use, my own home setup. Behind it, 15 years of shipping real systems to production.

Each dot is one WeatherXM station. The pulses are simulated.

How I work with AI.

AI isn’t a feature I add at the end. It’s how I think, design, write code and help run a company. Four layers, from the agent that holds WeatherXM’s knowledge to the setup my family uses at home.

OpenClaw, WeatherXM’s company mind

The company’s shared brain, in Slack

OpenClaw is an agent I helped create and keep evolving. It lives in Slack and runs on a shared company knowledge base; I recommended Open Knowledge Format as its specification and helped implement it. It analyzes support tickets and lead reports, discusses requirements with colleagues, takes features end to end, and has AWS and Cloudflare access for real infrastructure work.

  • Slack
  • Open Knowledge Format
  • Tickets & leads
  • Features end to end
  • AWS & Cloudflare

From a Slack thread to merged code

How a feature ships at WeatherXM

  1. Discuss

    OpenClaw clarifies the request with the people who asked for it.

  2. Design

    It works the solution through with a “CTO agent” against shared company context.

  3. Implement

    It writes the code and opens a pull request; CodeRabbit reviews it.

  4. Release

    Approval matched to risk, per repository and change.

Low-risk contentWebsite changes go straight to production.

Sensitive backend codeServices used by thousands of customers merge only after a person reviews the pull request.

Across 200+ repositories, software is now agent-led. People set direction, review and own the outcome. That includes me.

Products, not just internal tooling

  • Mobile AI assistant

    I proposed it and led two AWS AI hackathon teams; it was in production within a month. That work led to co-presenting the AWS for Startups keynote in October 2025.

  • First-line support

    I helped introduce a Dify agent grounded (RAG) in help-desk docs, past tickets and internal knowledge, with escalation to a person.

  • Found by AI

    I rebuilt exm.gr to be read by language models. New pilot enquiries came from customers who found Ex Machina through ChatGPT or Claude.

  • A small API for agents

    A weather-data API that AI agents can discover and pay for per request (x402), built end to end with AI-assisted coding and deployed automatically on Cloudflare.

Every day, at work and at home

I write code with Claude, Codex and opencode every day; this site and its CV were built that way. At home, I run Hermes Agent in a multi-agent, multi-user setup that my family uses.

  • Claude
  • Codex
  • opencode
  • Hermes Agent

Selected work

Weather-station networks, energy forecasting, motorway tunnels: real systems for real customers. They’re why I trust what I ship with AI now.

WeatherXM · real-time data · 2021–present

Data platform, from proof of concept to production

10,000 stations, streaming in real time.

I built the proof of concept and helped build the production platform that ingests real-time data streams from WeatherXM’s network of 10,000 weather stations: multi-instance ThingsBoard, Kafka and PostgreSQL. I built the first Android app and the client-facing API on top of it, and later, as Chief Product Officer, created the Product & R&D team and led WeatherXM Pro.

  • Kafka
  • ThingsBoard
  • PostgreSQL
  • Android
  • REST API
From proof of concept to a 10,000-station network. The data foundation behind WeatherXM’s apps, API and data products.

ADMIE · energy · 2016–2017

Built and run single-handedly

Data behind national grid balancing.

I independently designed, built and operated the weather-data backend used by Greece’s electricity transmission system operator to support balancing the national grid. I developed the Node.js services, messaging and document storage, provisioned the cloud infrastructure, and then migrated the system to Azure.

  • Node.js
  • Message queues
  • Document DB
  • Azure migration
Two years in production. Ownership from application code to the cloud environment that ran it.

Fysiko Aerio · energy · 2016–present

Technical delivery and ongoing customer ownership

A customer relationship built over a decade.

Since 2016, I have been the sole owner of the customer relationship and technical delivery for Fysiko Aerio. I built and maintain the weather-data services the company uses for gas and electricity demand forecasting in Athens and Thessaloniki, working directly with the customer as its requirements evolve.

  • Weather data
  • APIs
  • Demand forecasting
Ten years of continuity. Not just delivering a system, but understanding the customer’s needs and keeping it useful.

Olympia Odos · industrial IoT · 2019–present

Edge systems, field deployment, data analysis and machine vision

From a tunnel at midnight to edge AI.

For a jet-fan monitoring pilot, I built the RevPi gateway and cloud infrastructure: local time-series storage, on-device calculations and synchronization with ThingsBoard. I worked alongside colleagues installing and commissioning sensors overnight in motorway tunnels, then performed FFT-based vibration analysis and co-produced the customer report with a project partner.

On the follow-on smart-camera project, I helped build the edge-to-cloud pipeline: machine-vision algorithms running on NVIDIA edge devices, streaming detection events to our cloud platform.

Since 2019, the work with Olympia Odos has grown from smart tunnel monitoring into full situation awareness: edge cameras, weather stations, solar and wind farm production monitoring, pump monitoring and more. I have been hands-on throughout, leading the team that builds it and working with the customer in the field. Nobody called it forward-deployed engineering then, but that is what it was.

  • RevPi
  • ThingsBoard
  • Time-series
  • FFT analysis
  • Edge machine vision
The pilot led to a €200,000 project and helped establish the relationship and platform for a subsequent €300,000–€400,000 smart-camera situation-awareness project.

Across the whole product.

Job titles never quite fit, so here is what I actually do. Founding two companies meant owning whatever the product needed next; today I do all of it with AI alongside.

  • Product & UX

    Customer discovery, product strategy and UX design, grounded in an MSc in Human-Computer Interaction. Created WeatherXM’s Product & R&D team and led WeatherXM Pro.

  • Forward-deployed with customers

    Hands-on with the customer, on site and off: discovery, demos and pilots alongside the CEO, technical proposals, field installation and long-term ownership. In the field with Olympia Odos since 2019, and Fysiko Aerio’s sole owner since 2016.

    Integrations with Amazon, WooCommerce, an ERP and logistics partners

  • Mobile & web

    The first WeatherXM Android app, websites for WeatherXM and Ex Machina, and earlier 16 Android apps and a six-developer Android team.

    TypeScript / Node.js, Java / Android

  • Cloud & infrastructure

    Event-driven systems, messaging, caching, object storage, CI/CD and cross-cloud migration. Years as the person who kept the VMs, databases and pipelines running.

    AWS, Cloudflare, Azure

  • Data

    Real-time streaming and IoT ingestion at scale, edge gateways, data-lake querying and ETL, forecasting and signal analysis.

    Kafka, MQTT, PostgreSQL, MongoDB, Athena, Glue

  • Working with AI

    Agents that hold company knowledge and ship code, retrieval-grounded (RAG) assistants and risk-based human review. Used across a whole company and at home.

    Claude, Codex, opencode, OpenClaw, Hermes Agent, Dify

A builder’s path.

WeatherXM

Co-founder; WeatherXM was founded by the Ex Machina team. Head of Engineering, 2021–2023; Chief Product Officer, 2023–2025; Founder & Builder, 2025–present.

Helped create OpenClaw, the company’s Slack-based agent, and led the shift to agent-led software across 200+ repositories. Built the proof of concept and helped build the production data platform now streaming from 10,000 stations; built the first Android app and client-facing API. Created the Product & R&D team and led WeatherXM Pro’s product and design work.

Built a forecasting backtesting pipeline with a time-series foundation model (TimesFM) and evaluated ML weather models (GenCast, Aurora, AIFS ENS). Co-presented an AWS for Startups keynote in October 2025.

Ex Machina

Co-founder working across full-stack engineering, cloud infrastructure, UX and customer delivery. Built production data services, IoT pilots, edge machine-vision pipelines and integrations connecting commerce, ERP and logistics systems.

Led Meteoblue integration and cost-reduction work. Rebuilt exm.gr, generating pilot enquiries from customers who reported discovering EXM through ChatGPT or Claude.

Earlier foundations

Globo · Lead Android Developer · 2014–2015Hands-on enterprise-app development and leadership of six developers.

OpenCanvas · Co-founder, Android & Mobile UX · 2013SDK, cross-platform UX and automated white-label app builds.

NCSR Demokritos · Research Associate · 2011–2013Transport and security applications for EU research projects.

Mobzili · Android & UI, part-time · 2011–2013Designed and developed 16 Android applications.

Education

MSc, Human-Computer Interaction with ErgonomicsUniversity College London · 2010. Dissertation hosted by Sony Computer Entertainment Europe; awarded Distinction.

MEng, Electrical & Computer EngineeringAristotle University of Thessaloniki · 2008

Tell me what you’re building.

Athens, Greece

“Start a conversation” opens a contact form.