Clinical reasoning
Translate pharmacovigilance, longitudinal observation and safety requirements into concrete data and interface decisions.
Medicine × software engineering × evidence
Clinical insight.
Production-minded software.
Physician, Clinical Pharmacology Resident and eHealth Developer in Madrid. I turn healthcare questions into inspectable prototypes, document their limits and improve them through real deployment feedback.
3 public codebasesPortfolio, clinical research POC and mail platform.
Safety firstLimits and non-clinical boundaries are part of the product.
Full-stack deliveryReact, Node, Python, Supabase, Render and GitHub Actions.
Versioned learningRelease notes preserve how each system evolved.
A clinician who ships
My work starts with a clinical or operational problem, narrows it to a testable software boundary, and keeps uncertainty visible. The result is not a claim of medical validation: it is evidence of product thinking, engineering iteration and responsible scope.
PUBLIC PORTFOLIO // CODE + DECISIONS + LIMITS
Software delivery
Build across frontend, backend, database and deployment layers, then use production discrepancies to drive the next version.
Responsible communication
Separate prototypes from validated products, distinguish signals from causality and document what a system cannot safely do.
Selected work
Three projects show an evolution from learning exercises to systems with explicit deployment, security and domain constraints.
STATUS LABELS DESCRIBE THE CURRENT PUBLIC VERSION
Ψ
PsychApp v0.2.0
A research POC for clinician-supervised longitudinal modelling using synthetic data, coordinated agent contracts, inspectable uncertainty and a safety layer. It is not a medical device, diagnostic service or crisis tool.
React + ViteNode.jsSupabase schemaPython model serviceSynthetic data
@
Domain Mail v1.2.0
A Flask webmail interface that progressed from a Python object model to real IMAP/SMTP integration and a self-hosting architecture. The public Render deployment now uses a synthetic showcase mode, so visitors can evaluate the interaction without exposing mail credentials.
Python + FlaskIMAP / SMTPGunicornRenderDocker mailserver design
PV
Pharmacovigilance API Workbench
A browser workbench separating restricted European pharmacovigilance access from lawful public aggregate analysis. It supports local CSV exploration, public openFDA queries and exploratory ROR/PRR calculations without presenting signals as incidence or causality.
Vanilla JavaScriptSupabase Edge FunctionsopenFDAROR / PRRAuthorised proxy boundary
How I work
The portfolio is intentionally explicit about trade-offs. For healthcare software, a polished interface is useful only when the data path, operational state and clinical boundary can also be inspected.
1. Define the clinical boundaryState intended users, excluded uses and escalation points before adding automation.
2. Build a narrow vertical sliceConnect interface, API, data and deployment early enough to expose integration failures.
3. Treat production as evidenceCompare repository intent with the live service, then fix configuration drift instead of hiding it.
4. Version the reasoningKeep release notes and retrospective posts so an employer can inspect growth, not only the final screenshot.
ClinicalSafety framing, pharmacovigilance and human review.
ProductScope, modes, workflows and honest status communication.
EngineeringAPIs, frontend, data models, tests and deployment.
SecuritySecrets, origin boundaries, RLS awareness and safe demos.
Versioned evolution
Milestones are tied to repository history. Retrospectives preserve the date of the original milestone while clearly stating when the retrospective itself was published.
Portfolio v2.0 + cross-project audit
Reframed the site as an employer-facing clinical engineering portfolio, consolidated documentation and hardened public endpoints.
PsychApp v0.2 research POC
Rebuilt the project around clinician supervision, population modelling, synthetic demonstration data, safety controls and reproducible research assets.
Domain Mail v1.1 performance pass
Replaced N+1 IMAP fetches with batching and documented the transition from classroom model to operational mail architecture.
Domain Mail v1.0 architecture
Added the self-hosted mail appliance design, TLS/DNS operations and a real IMAP/SMTP web interface.
Pharmacovigilance workbench
Published the public aggregate workflow and separated it from restricted EudraVigilance access.
PsychApp v0.1 first prototype
The original PsychDeep note remains unchanged; a new retrospective maps that first concept to the later research architecture.
Python Email System v0
A small object-oriented exercise established the message, inbox and user model that later evolved into Domain Mail.
Public CV // evidence-linked
A concise profile for hiring conversations
Review the clinical–technical profile, demonstrated stack, selected projects and current product boundaries in a print-friendly page.
Writing & release notes
Technical notes connect visible features to architecture decisions, limitations and what changed between versions.
PsychApp v0.2
The research POC rebuild, clinician-supervised architecture and the gap between repository and live deployment.
Domain Mail v1.2
Why the public deployment became a synthetic showcase and how live IMAP/SMTP remains isolated.
Portfolio v2.0
Documentation consolidation, security fixes and a clearer employer-facing narrative.
Contact // Madrid, European Union
Let’s discuss clinical software
For roles, collaborations or technical conversations about digital health, clinical pharmacology and responsible product development.