I work on AI systems where abstractions start to fail, production agents, evaluation pipelines, and AI-first SaaS platforms that have to operate under real world constraints.
My background spans software engineering, applied ML, and system design. These days I operate as an independent consultant, partnering with startups and research driven teams on agentic architectures, reliability, and production readiness.
I write long-form essays at Rooted Layers, exploring the layers beneath modern AI systems from hardware–software codesign to autonomous agents.
Links:
Independent AI Systems Consultant working on production AI, agentic systems, and system architecture beyond the prototype stage.
I focus on evaluation, reliability, and failure modes in AI-first SaaS systems, bridging research ideas and real-world deployment.
- AI systems architecture & agentic systems
- Evaluation, reliability, and governance
- Technical advisory & selective consulting
Pinned repositories below reflect current and representative work.
Over the years I’ve worked across software engineering, applied ML, and AI systems. Some earlier projects and explorations are archived but still useful as context:
- Early ML & data systems experiments
- Infrastructure and backend engineering projects
- Research-oriented prototypes and tooling
(Older work is kept for reference and historical context.)
-
A presentation at ICLR2020 of a lightning talk for the CyberSecurity social event, on NLP+CyberSecurity.
-
Some NLP notebooks and other problems I liked, especially on Alzheimer's text classification.
-
Towards a production grade Flask server to serve TensorFlow models, written in 5 days flat, which means it needs some more work. :)
-
A showcase Data Science project, predicting taxi ride demand for the city of Lima.
-
My Thesis with code, for my MSc. in CS, on multicore MapReduce relational operators, for the University of Edinburgh.
-
A template I use, to create my resume with Latex.