Rafael – Webhooks, Workflow Automation, SQL

Digital marketing automation platform focused on lead generation workflows for professional services clients. The system integrates advertising platforms, CRM systems, and messaging channels to automate the full lead lifecycle - from ad click and lead capture through CRM entry to WhatsApp and email follow-ups. The platform relies on 20+ automated workflows built using tools such as n8n, Make, and Google Apps Script, with API-driven integrations ensuring seamless data flow between systems.

• Developed automation workflows using n8n and Make to integrate systems, orchestrate data flows, and reduce manual operational work;
• Built integrations between business tools using REST APIs and webhooks, automating data synchronization between CRMs, spreadsheets, and internal systems;
• Implemented scripts with Google Apps Script and Clasp to extend Google Workspace functionality, automating workflows in Google Sheets, Gmail, and Google Docs;
• Designed and structured automation workflows with a focus on maintainability, clarity, and scalability, enabling long-term evolution of automated processes;
• Implemented error handling, logging, validations, and retry mechanisms to improve reliability and reduce operational failures;
• Optimized automation flows to improve execution time, resource usage, and stability, particularly in data-intensive processes;
• Documented automations and integrations to support maintenance, troubleshooting, and future development;
• Developed supporting scripts and small internal tools to assist automation workflows and solve specific integration needs.

Web-based platform designed to simplify annual income tax filing for stock investors, reducing the data entry process from hours to under 10 minutes. The system focuses on automating and streamlining tax reporting workflows through an intuitive user experience and structured data handling. Led full-stack development with a strong focus on front-end architecture, UX/UI design, and product development, while also contributing to the design of tax calculation logic based on domain knowledge in accounting.

• Built a full-stack web platform from scratch to simplify stock tax reporting, reducing data entry time from hours to under 10 minutes;
• Delivered core product features including authentication, paywall integration, IR reporting, portfolio analytics, and document upload workflows;
• Led full-stack development with a strong focus on front-end architecture, UX/UI design, and product development;
• Modeled and implemented tax calculation logic based on accounting domain knowledge
• Developed the front-end using Vite + React, the back-end using FastAPI, and managed data storage with Supabase (PostgreSQL);
• Continuously improved user experience, fixed bugs, and strengthened system security and reliability over time.

Automated lead operations platform designed for a marketing organization managing paid and organic lead generation across multiple clients. The system connects lead capture forms, marketing channels, and client CRM systems into a unified workflow, enabling end-to-end automation of lead processing. Each incoming lead is automatically routed from the point of submission through enrichment, tracking, and CRM synchronization, ensuring seamless handoff to sales teams and first-contact initiation without manual intervention. The platform improves operational efficiency by eliminating fragmented processes and centralizing lead lifecycle management across multiple systems.

• Designed and built an end-to-end, event-driven lead automation pipeline serving multiple clients within a single unified workflow;
• Implemented real-time lead capture with instant team alerts, optimized for a sub-2-minute response time requirement;
• Developed automated data enrichment, including timestamp parsing, phone normalization with click-to-message link generation, and day-of-week derivation;
• Built marketing attribution parsing to extract Google Ads parameters (campaign, ad group, creative, keyword, match type, device, network) and resolve campaign IDs into readable names;
• Implemented automatic lead-source classification (paid, organic, referral, and others);
• Created dynamic multi-client routing logic that determines each lead’s destination CRM at runtime via a central configuration sheet;
• Developed deduplication logic matching contacts by phone number, with conditional create-or-update CRM record handling;
• Automated CRM record creation with extensive field mapping, including pipeline stage assignment and triggering of first-contact workflows;
• Engineered system reliability features, including retry with backoff, rate limiting, error recovery fallbacks, and a manual lead recovery mechanism.

This project was built for a marketing agency that manages Google Ads campaigns for a portfolio of clients. It's an automated budget-monitoring and alerting system that keeps track of every client's ad account balance throughout the day and warns both the agency and the clients - over WhatsApp and email - before an account runs low enough to put campaigns at risk. The goal was to make sure no client's ads ever stop running for lack of funds, without anyone having to check balances by hand.

• Built a scheduled automation that monitors a portfolio of Google Ads accounts multiple times per day on weekdays;
• Integrated with the Google Ads API to retrieve the full client account list and automatically filter for active accounts;
• Developed per-account budget logic to calculate remaining balances and distinguish between prepaid and postpaid billing models;
• Automated write-back of current balances and payment types to a central control spreadsheet while preserving existing formatting;
• Designed a two-tier alerting system, sending recharge reminders when accounts drop below 25% of budget and escalating urgent warnings below 10%;
• Implemented payment-type-aware alert routing, adjusting notification timing for prepaid and postpaid clients;
• Automated personalized client alerts via WhatsApp and email, including calculated top-up recommendations for each account;
• Built parallel internal notifications to alert the agency team when client accounts require attention;
• Engineered reliability and anti-ban safeguards, including API retries, randomized send delays, deduplication, and graceful handling of missing or unmatched records;
• Architected the solution as two linked workflows, where the monitoring process automatically triggers the alerting workflow.

This project was built for a marketing agency that produces SEO-optimized blog content for its clients. It's an AI-powered content pipeline that turns a raw transcript (for example, from a recorded interview or video) into a fully drafted, structured, SEO-ready blog article saved straight to the team's Google Drive - collapsing what used to be hours of writing and formatting into a hands-off, automated process.

• Designed and implemented an end-to-end AI content generation pipeline that converts long-form transcripts into publish-ready blog articles;
• Integrated large language models (OpenAI Assistants and Google Gemini) using an agent-based architecture for structured content planning and generation;
• Built a two-stage generation workflow where an AI agent first creates a structured article outline (title, SEO slug, meta description, keywords, and section plan), followed by section-by-section content generation;
• Engineered conversational memory between sections to ensure coherent transitions and avoid content repetition across the article;
• Enforced SEO and editorial constraints programmatically, including keyword placement, word-count targets, image placement guidance, and controlled calls-to-action;
• Implemented brand voice guardrails to ensure outputs remained faithful to the source transcript and avoided introducing off-topic or fabricated content;
• Automated Google Workspace integration, including Drive folder creation and structured Google Docs generation with section-level writing;
• Logged all pipeline runs to a central control spreadsheet, including output links, source references, and timestamps for full traceability;
• Built flexible input handling to support both Google Docs URLs and raw text inputs with automatic routing;
• Implemented API rate limiting, batching, and pacing mechanisms to ensure stability across multi-step generation workflows.

A web development project focused on building and maintaining multiple client-facing websites across different industries, with an emphasis on performance, SEO, and scalability. The platforms were built using Next.js and included marketing and content-driven websites optimized through SSR/SSG, performance enhancements, and modern frontend practices. The work also involved integrating third-party APIs for extended functionality and setting up CI/CD pipelines with automated deployments via Vercel to ensure efficient and reliable releases.

• Built optimized landing pages and full-featured web applications with authentication systems and database search powered by third-party APIs;
• Created automations using Python, Selenium, and n8n, helping to streamline internal processes.

This project was built for a marketing agency that runs its internal operations and client work. It's a project-management automation that adds custom one-click actions to tasks: with a single button, a team member can move a task into the current or next sprint, recurring tasks can be regenerated automatically, and tasks can be shuffled between sprints - all of it handled behind the scenes so the team manages work by clicking instead of doing manual sprint housekeeping.

• Built an event-driven automation triggered by custom action buttons on ClickUp tasks, responding to user interactions in real time;
• Developed sprint detection logic that reads sprint date ranges from list names and automatically identifies current and upcoming sprints based on the current date;
• Implemented multiple one-click task actions, including adding tasks to the current sprint, adding them to the next sprint, and moving tasks between sprints in a single operation;
• Automated recurring task generation by calculating next occurrences from due dates, mapping them to the correct day of the week, and recreating tasks with pre-filled scheduling fields;
• Built logic to read and update ClickUp custom fields via the ClickUp API, including resetting and toggling button states after actions were executed;
• Implemented a routing layer to detect which button was clicked and dispatch the corresponding action, with safe no-op handling for unsupported cases;
• Handled date parsing, day-of-week mapping, and timezone-aware scheduling within custom automation logic.