01
Pin down the outcome
I clarify the user, the workflow, the constraints and the result that would make the project worth shipping.
I build your software solution end-to-end.
Frontend, backend, AI, whatever you need. I turn your ideas into shipped software and the operational details around them.
Available · Q2 2026 · Buenos Aires · Remote
Ways I can help
Four clear ways to turn business problems into working software.
Focused MVPs across AI workflows, full-stack product work, internal tools and automations.
// ai-powered
AI workflows
LLM integrations, agents, MCP-style tool use and reviewable AI workflows embedded inside practical products.
LLM APIsMCPRAG-ready flowshuman review
Full-stack web products
Features that need real state, forms, APIs, databases, roles, validation and deployment context.
Next.jsNode/NestJSPostgreSQLauth flows
Internal tools and dashboards
Operational screens for teams that need to see, filter, approve and act on their data without fighting spreadsheets.
tableschartsadmin panelsreports
Automations and integrations
Workflows that connect tools, remove repeated handoffs and make the next action traceable.
n8nwebhooksjobsnotifications
Selected builds
Proof across product, automation, AI and frontend work.
A focused set of portfolio projects covering the freelance work I want more of: AI systems, internal tools, product infrastructure and polished web experiences.
Applied AI for operations
AI WhatsApp Operations Assistant
// 01Flagship project
A WhatsApp-first assistant for service businesses that answers customers, remembers business rules and supports operators with practical tools.
Next.jsNestJSWhatsApp APIPostgreSQLLLM API
context
Small businesses often run sales and operations through chat, but the knowledge lives in scattered messages, notes and repeated decisions.
architecture
Conversational AI bounded by business tools, memory, human review, cost calculation and observability around the workflow.
project
Next.js and NestJS product surface with WhatsApp integration, business CRUD, memory, cost calculator and reviewable AI responses.
outcome
A flagship case study for practical AI: automation that helps the business without hiding control from the people running it.
Internal developer tooling
Infra Orchestrator + MCP
// 02Portfolio project
A small orchestrator for local services and AI-agent tooling, designed to make developer workflows easier to start, inspect and control.
TypeScriptMCPNode.jsReactCLI tooling
context
Multi-service projects get slow when setup, logs and commands are scattered across terminals and docs.
architecture
Typed command layer, MCP wrapper, service status model and a UI that exposes only the actions a developer actually needs.
project
A dashboard and tool wrapper that starts services, shows health, exposes logs and lets AI agents call bounded infrastructure actions.
outcome
A compact internal-tool case study for orchestration, MCP integration and practical developer automation.
LLM observability
Mini-LangSmith Self-Hosted
// 03Portfolio project
A self-hosted tracing and evaluation tool for teams that want visibility into prompts, runs and regressions without a heavy platform.
Next.jsPython SDKPostgreSQLLLM evalsDocker
context
AI workflows are hard to improve when prompts, traces, metrics and failures are spread across logs and screenshots.
architecture
Trace ingestion, run comparison, baseline metrics and a small SDK that keeps instrumentation explicit.
project
Dashboard, API and Python SDK for capturing LLM runs, reviewing outputs and comparing changes over time.
outcome
A focused observability project that shows product thinking around AI quality, not just model calls.
Polished web experience
Service Brand Landing
// 04Portfolio project
A motion-heavy landing page for a service brand, built to show warm visual direction outside the usual technical SaaS look.
Next.jsTypeScriptTailwind CSSmotionCloudflare Pages
context
Creative businesses need a site that feels like the product before the visitor reads every detail.
architecture
Static-exported Next.js page with localized copy, responsive motion, optimized assets and clear conversion paths.
project
Hero, product sections, booking path and brand-led interactions tuned for screenshots, mobile and fast loading.
outcome
A visual case study for polished frontend work, brand expression and conversion-oriented page design.
AI knowledge systems
Agentic RAG Starter
// 05Portfolio project
A starter project for retrieval workflows with agents, evaluation harnesses and honest documentation about tradeoffs.
Next.jsTypeScriptRAGLLM APIeval harness
context
Teams want chat over their knowledge base, but the useful part is retrieval quality, traceability and failure handling.
architecture
Ingestion pipeline, retrieval layer, tool-using agent, eval baseline and comparison docs for realistic expectations.
project
Demo app with bounded agent actions, document retrieval, eval harness and public README explaining where RAG helps and where it does not.
outcome
A reusable reference for AI workflow architecture with enough rigor to discuss quality, not just demos.
Capabilities
The stack is broad, but the offer is simple: ship the right system.
These are SOME of the tools and practices I can bring together depending on what the client actually needs.
Frontend
ReactNext.jsTypeScriptTailwind CSSUI systems
Backend
Node.jsNestJSREST APIsPostgreSQLRedis
AI & Automation
LLM integrationsMCP serversAI agentsRAG workflowsn8nhuman review loops
Architecture
clean architecturemodular systemsAPI designtyped contractsrefactoring
Tools & Delivery
GitGitHubDockerCloudflare Pages
AI Coding Tools
Claude CodeCodexopencodePi DevWindsurfCursor
Process
A practical path from problem to shipped MVP.
I keep a direct process: I define the objective, outline the moving parts, launch the essential version, and iterate based on real usage.
implementation path
Each step reduces ambiguity until there is a working MVP with clear ownership and reviewable behavior.
02
Map the workflow
I trace screens, data, integrations, permissions and repeated manual steps before choosing what to build.
03
Scope the first MVP
I choose the smallest useful version that proves the behavior without pretending the future is fully known.
04
Build the core
I implement the UI, backend, validation, integrations and AI workflows needed for the workflow to actually run.
05
Ship and tighten
I deploy, check the important paths, make the actions reviewable and iterate from what real usage shows.
About
Full-stack engineer. Architecture, performance, security — not just features.
I combine frontend work, backend implementation, integration work and AI/automations direction.
Buenos Aires · remote-ready
Buenos Aires-based, working remote. I've shipped React/TypeScript at enterprise scale — microfrontends, shared packages, real review cycles. Backend side: Node, NestJS, databases, the usual stack.
Now I'm focused on applied AI — not the buzzword, the practical kind. If a workflow gains from extraction, agents or automation, I design that layer into the product.
When I design a system, I think about how it holds up under real use — not just the happy path. Boundaries that don't leak, performance that doesn't degrade, security that isn't bolted on later, and architecture that stays clean as the product grows.
Frontend that feels intentional, not assembled
Backend and integrations built for real load
Architecture decisions that scale beyond the first version
AI and automation used where they earn their place
full-stackarchitectureperformancesecurityAIautomation
How I think about software
The right software makes complex work easier to move through.
I care about useful systems: interfaces that explain themselves, backend boundaries that are hard to misuse, and automation that reduces work without hiding control.
Bringmeagoalandconstraints.I'llfigureouttherest.
Frontend,backend,automations,AIworkflows,integrations—Iworkacrossthestack.Newtechonlygoesinifitearnsitsplace.
Idon'tleavethingshalf-done.Ibuildwhat'sneeded,andstayuntilitworks.
Open for projects
Have something to build?
Tell me what you are trying to build. If you need software, systems or AI, we can create the right solution.