Selected Work

Building products
that work at scale.

I'm Marcos Coelho — a technical founder and full-stack developer. I design, architect and ship end-to-end systems.

Available for new projects·hello@marcoscoelho.pt
bash — zsh

Selected Projects

01—LIVE
2025
Orbit Analytics dashboard screenshot
Open case study

Multi-tenant SaaS analytics platform

Orbit Analytics

A full-stack analytics platform processing 50k+ events per day with real-time dashboards, custom funnel analysis, and automated alerting. Built for indie SaaS founders who need Mixpanel-level insights without the enterprise pricing.

50k+events / day
Next.js 16SupabaseTypeScriptRechartsVercelPostgreSQL
SQL
SELECT
  event_type,
  COUNT(*)            AS total,
  DATE_TRUNC('day', created_at) AS day
FROM events
WHERE workspace_id = $1
  AND created_at > NOW() - INTERVAL '30 days'
GROUP BY 1, 3
ORDER BY 3 DESC;
View case study
02—LIVE
2025
Pulso AI workflow automation interface
Open case study

AI-powered content automation engine

Pulso AI

An end-to-end AI workflow engine that ingests raw content signals, classifies intent with a fine-tuned model, generates structured outputs, and distributes them across multiple channels simultaneously.

80%less manual work
PythonFastAPIn8nOpenAI APISupabaseDocker
Python
async def process_content(payload: ContentPayload):
    classified = await llm.classify(payload.raw)
    structured = await llm.extract(classified)

    await asyncio.gather(
        publisher.slack(structured),
        publisher.newsletter(structured),
        db.store(structured)
    )
View case study
03—LIVE
2026
marcoscoelho.pt site screenshot
Open case study

Personal site, blog and portfolio

marcoscoelho.pt

Designed and built from the ground up as a living technical system. Headless CMS with real-time preview, edge-cached content delivery, structured metadata, and a clean content pipeline from Sanity Studio to production in under 3 seconds.

<3spublish to edge
Next.js 16Sanity v5SupabaseTailwind CSS 4VercelTypeScript
TypeScript
export const client = createClient({
  projectId: process.env.SANITY_PROJECT_ID,
  dataset: 'production',
  perspective: 'previewDrafts',
  useCdn: false,
  token: process.env.SANITY_READ_TOKEN,
})
View case study

Technical Capabilities

The full stack,
end to end.

Full-Stack Engineering

End-to-end product delivery — from schema design to polished UI. React, Next.js, TypeScript.

API Architecture

RESTful and event-driven APIs built for scale. FastAPI, Node.js, edge functions, webhooks.

AI / LLM Integration

Production LLM pipelines, prompt engineering, RAG systems, and fine-tuning workflows.

Workflow Automation

n8n, custom Python scripts, and AI agents that eliminate repetitive operational work.

Data Engineering

Postgres, Supabase, schema design, real-time subscriptions, and analytical query optimisation.

DevOps & Deployment

Vercel, Docker, CI/CD pipelines, environment management, and zero-downtime releases.

Performance Engineering

Core Web Vitals, edge caching, bundle optimisation, and database query profiling.

Systems Architecture

Designing modular, maintainable systems. Monorepos, bounded contexts, clean separation of concerns.

CMS & Content Pipelines

Sanity, headless CMS integrations, structured content modeling, and Portable Text.

Product Analytics

Instrumentation, funnel analysis, event tracking, and dashboard design for data-informed iteration.

Process

How I work.

01

Discover & Scope

I start by mapping the real problem — not the stated one. User interviews, system audits, and constraint analysis before a single line is written.

Outcome: a clear problem statement, technical constraints, and a prioritised feature scope.

02

Architecture First

I design the data model and system boundaries before touching the UI. Schema, API contracts, and service interfaces are defined up front.

Outcome: an ERD, API spec, and component tree that acts as the source of truth for the build.

03

Build & Iterate

Short cycles. Shipping working software early, integrating real feedback, and adjusting the plan. No big-bang releases.

Outcome: a production-ready increment every 1–2 weeks, continuously deployed to staging.

04

Ship & Monitor

Deployment is not the finish line. Observability, error tracking, analytics, and performance baselines are wired in before launch.

Outcome: a monitored, documented system that can be handed off or extended without tribal knowledge.

Typical stack topology


  ┌──────────────────────────────────────────────────────────────┐
  │                        Client (Browser)                      │
  │                    Next.js App Router + React 19             │
  └────────────────────────────┬─────────────────────────────────┘
                               │ HTTPS / RSC
  ┌────────────────────────────▼─────────────────────────────────┐
  │                     Edge Network (Vercel)                     │
  │          ISR cache · CDN · Middleware · Edge Functions        │
  └────────┬───────────────────────────────────────┬─────────────┘
           │ Server Actions / API Routes            │ Webhooks
  ┌────────▼────────────┐             ┌─────────────▼───────────┐
  │   Sanity Studio     │             │   Automation Layer      │
  │   (Headless CMS)    │             │   n8n · FastAPI · cron  │
  │   Portable Text     │             │   LLM Pipelines         │
  └─────────────────────┘             └─────────────────────────┘
                               │ Postgres / Realtime
  ┌────────────────────────────▼─────────────────────────────────┐
  │                        Supabase                               │
  │       Auth · Postgres RLS · Storage · Realtime               │
  └──────────────────────────────────────────────────────────────┘

Contact

Let's build
something.

Open to consulting, technical co-founder conversations, or interesting full-stack / AI engineering projects.

hello@marcoscoelho.pt
GitHubLinkedInTwitter / X