About
I build software to understand problems.
Nicolás Bruna Software Developer
I build software around real problems I run into, experiment with technologies along the way, and keep coming back to the same intersection: software, AI and systems.
About
I am Nicolás Bruna, a software developer from Chile. I studied Ingeniería Informática at INACAP Valparaíso and obtained the Analista Programador qualification.
Most of what I know about software, AI and systems comes from building things: local AI tooling, developer tools, product experiments and full-stack applications. I learn by turning a problem into something executable, investigating why it behaves the way it does, and iterating until the system underneath makes sense.
This portfolio is a record of that process — the problems I chose to investigate, the software that came out of them, and what each project taught me.
How I work
Four principles that describe how I actually build, not a methodology I sell.
Build
I learn by turning an idea or a problem into something executable — a script, a tool, a prototype — instead of studying it from a distance.
Investigate
When something does not work, I want to understand why rather than simply bypassing it. The failure is usually where the learning is.
Experiment
I use projects as a way to test technologies, architectures and ideas against real constraints instead of comparing them in the abstract.
Iterate
The first implementation is rarely the final one. I improve systems as I understand them better, and I am comfortable rewriting what I got wrong.
What I build
Areas of interest and current focus — directions I work in, not mastery I claim.
- Software
- Full-stack and backend work: applications, APIs and developer tooling built around concrete problems.
- AI
- Local AI, model runtimes, inference and AI-assisted development — especially the layer between models and the people running them.
- Systems
- Linux, infrastructure, networking and the interaction between hardware and software — the environment the code actually runs in.
- Architecture
- Separation of concerns, abstractions, compatibility decisions and execution pipelines that stay maintainable as they grow.
These are areas I work in and keep learning about — not a claim of expertise in any of them.
Evidence, not adjectives
The three projects in this portfolio, and what each one says about how I work.
- CastleArqLocal AI orchestration platformStarted from a real problem with local AI setup — and turned into architecture work.
- SnapContextAI context management for developersPersonal friction with AI-assisted development, turned into reusable tooling.
- ProdLocalesPaused — project statusLocal product pricing and analyticsA product experiment that taught its lesson — and was deliberately paused.
Each project has a full case study describing what was built, what was decided and what was learned.
Current focus
Where my attention is right now. Not equally advanced — just the direction I am moving in.
- software architecture
- local AI
- AI-assisted development
- Linux
- backend and full-stack development
- infrastructure and systems
- a stronger professional portfolio
Learning by building
Curiosity is useful when it leads to experimentation.
My projects usually start before I have all the necessary knowledge: a problem appears, I investigate possible solutions, I build something, I discover its limitations, and my understanding gets sharper with each round.
That means early versions are often wrong in instructive ways. I treat that as part of the process rather than something to hide — the portfolio keeps the reasoning, not just the result.
I build software to solve problems I encounter, then use the process to understand the system underneath.
Education
- Ingeniería Informática
- INACAP — Valparaíso, Chile
- Analista Programador
- INACAP
What I build matters more here than where I studied — this page keeps education compact on purpose.
Technologies
Tools that appear in the projects and in this portfolio itself. No ratings, no levels — just what the work uses.
- Languages
- Python
- TypeScript
- JavaScript
- SQL
- Development
- Astro
- VS Code
- Git
- AI / local AI
- Ollama
- llama.cpp
- local model runtimes
- AI-assisted development tooling
- Systems
- Linux
- Git
Docker appears only as the sandbox SnapContext uses for risky commands.
Anything else I touch is either experimental or aspirational — it does not belong in this list yet.
Professional direction
My career is still developing, so I keep this open on purpose. I am interested in work where I can keep building and keep learning — especially where software meets real technical constraints.
- software development
- backend / full-stack
- AI and software tooling
- systems and infrastructure
- solving technical problems
Still building. Still learning.
This portfolio is a record of the problems I have chosen to investigate, the software I have built, and the systems I am still learning to understand.