freyama.de - The homepage of Freya Madeline Broderick.
Built to resemble a terminal, it is fully interactive and contains all of the usual things a personal website would contain, plus maybe some easter eggs!
For a list of available commands, run `ls /bin`.
For more information, run `help help`.
Github - https://github.com/freyamade/fresh.
freya madeline broderick
full stack developer, platform / devops engineer, problem solver, puzzle lover.
fluent in python & typescript. comfortable with go, ruby, rust.
love terraform, django, vuejs. experienced builder of rest monoliths and microservices, both on gcp and bare metal.
for more in-depth information, i have also ma.de/a/cv
https://freyama.de
/a/cv.pdf - My fullform CV containing this site's details in a more easily accessible fashion.
/some-slides - Collection of all slides for brief tech-talks given during my time in UCC Netsoc.
# fresh
- Version 2 of the Terminal that powers [my website](https://freyama.de).
- Built from the ground up using Vue3 and Pinia, with some small libraries for some helpful features;
- [node-path](https://github.com/chaiNNer-org/node-path)
- Polyfill for nodejs path library, used for managing file system parsing
- [option-parser](https://github.com/tests-always-included/option-parser-js)
- For parsing flags and arguments for commands that have flags available
- [@highlightjs/vue-plugin](https://github.com/highlightjs/vue-plugin)
- For syntax highlighting files using `cat` for some easier reading
- Works like a real terminal, with a custom-built file system module.
- Uses vite glob importing to read files, and commands are correctly contained within the `/bin` directory, and are loaded on demand when running the command for the first time.
## noJs
If you don't want to use javascript, you can still see the outputs of some commands at https://freyama.de/noJs.html
This output is generated at build time via running the commands and pre-rendering using [Lightpanda](https://lightpanda.io) so the information is always up to date with the full terminal experience!# Honu.ai
## Lead Backend Developer
### September 2023 - March 2026
- Architected a zero-trust credentials proxy system allowing AI agents to securely interact with third-party APIs (e.g., Google Analytics) without exposing raw credentials.
- Primary backend owner for the Decision Infrastructure, balancing high-security requirements with seamless agent integration across GCP services.
- Managed the building and rollout of microservices in Python, using FastAPI, onto Google Cloud Platform services like Cloud Run, or Compute Engine, using Terraform and Terragrunt.
- Helped develop, and managed, a contextual programming system for the sharing of Business Context information, which was the underlying system of the Honu Decision Infrastructure, to power decision making in context for AI Agents.
- Developed a Data Proxy system where Agents could create Data Stores in our system, using Bigquery initially but was extensible for any backend in future, and have them accessible by any other Agents in the User’s Contextual Model.
- Built a custom task-based API using FastAPI, Google Cloud Task Queues, and Redis, for running LLM powered pipelines to turn User ideas into full Business Plan for the generation of new SaaS-based SMEs.
- Created a custom Chat infrastructure for the Decision Infrastructure to allow Users and Agents to converse with the Business Context in mind, by sending messages with Business Artefacts linked to establish context for the Agent.
- Built multiple data pipelines for the ingestion of user’s analytics and metrics data from various systems into the platform for the Agents to analyse and use to make decisions.
- Initially using Airflow DAGs running on Google Cloud Composer, but eventually migrated to a task system running on the platform directly, leveraging the Credentials and Data proxy systems to gather data and make it natively accessible to all Agents on the system.