Backend incantations,made to last.
I'm Oana. 13 years writing software — 7 of them keeping a HIPAA-compliant healthcare platform alight for clients ranging from solo practices to WebMD; now on Azure cloud-and-IoT systems for a global animal-health company via Toptal. Along the way: mobile games, multiplayer netcode, and a Danish cat rescue's entire web platform. I work AI-augmented — agentic coding with Claude Code, Copilot and Codex is my daily craft, and I build AI into the products I touch. If you have a project — or simply think we might work well together — get in touch. I genuinely enjoy burning the tricky bugs.
Spells from the tome.
A pair of long-running backend projects I'm proud of. Tap a card to peek beneath the seal.
HIA Studio - Healthcare Content Management Platform
HIPAA-compliant content management system for healthcare organizations to create AI-powered patient education presentations.
HIA Topics - AI-Powered Patient Education Platform
Enterprise healthcare platform delivering AI-powered patient education through physician-authored interactive presentations.
What I do for fun.
Side projects that quietly sharpen the day-job — a cat rescue's entire web platform and a fully AI-automated history-song factory.
Kattekøbing Members Portal - Cat Rescue Management System
Full-stack GDPR-compliant membership portal for a local cat rescue — 275+ members, secure administration, automated communication, an email-enabled forum, and AI-assisted member-data processing.
Decebal Cântă — AI-automated history songs for Romanian high-school students
A history-revision tool I'd have loved as a teenager: Romanian historical figures singing their own stories across metal, pop, trap, and manele. Two AI agents — Claude and Codex — run most of the pipeline, from cross-checked lyrics and synced lyric videos to portrait Shorts, auto-uploaded YouTube content, and matching quizzes. Free for students cramming for the history baccalaureate.
Backend, with an AI familiar.
Agentic coding is my daily craft — and I build AI into the products I work on, with production discipline.
Agentic by default
I orchestrate Claude Code, GitHub Copilot and Codex — not autocomplete. I also build tiered agent pipelines with parallel triage, provenance-aware extraction and independent verification, used across 1,500+ source records. I own the architecture and review.
AI, built into your product
LLM features, structured extraction and automation pipelines — wired into real backends with production guardrails, informed by years of integrating AI, voice, and data services into healthcare systems.
AI workflows, debugged & tuned
Flaky agents, runaway token costs, prompts and skills that quietly confuse the model — I benchmark coding agents against real specifications, dig into why workflows misbehave, and make them reliable.
Engineers keep asking
I've become the one teams pull in to show their other departments how to actually use AI well. I love that part — helping people get more out of these tools is my favourite kind of nerding out.
How I work.
A few principles I bring to teams, systems, and projects.
Find the real problem
I ask questions, inspect the system, and make sure we're solving the right thing before committing to a direction.
Make the work explicit
Scope, assumptions, trade-offs, and unknowns should be visible—not trapped in someone's head.
Ship and verify
I prefer small, testable slices with the strongest practical checks available before calling work complete.
Leave systems better
Clearer architecture, useful tooling, maintainable code, and shared knowledge should remain after the immediate problem is solved.