Software Engineer
Galidor Research
— Present
Description
Backend systems, business automation, AI workflows and production infrastructure for internal operations and customer-facing services.
My contribution
Independently design, develop, optimize, and operate backend systems, automation platforms, AI workflows, and production infrastructure for internal operations and customer-facing services.
Licitor — High-Performance Auction Automation
Problem
Public auctions without explicit APIs required time-sensitive bidding decisions.
My contribution
Developed and optimized RPA combining visual browser automation, computer vision, and specialized numeric recognition. Built the system to handle hundreds of concurrent bids across multiple customers, supporting high-throughput, time-sensitive operations. Technology stack: Go 1.25, Gin, GORM, PostgreSQL, Vue 3, TypeScript, Pinia, Vite, Python, FastAPI, OpenCV, Tesseract/tesserocr, Firefox, Xvfb, x11vnc, websockify/noVNC, xdotool, Docker Compose, and Nginx.
Results
Replaced generic OCR-based screen reading, which took approximately 5–10 seconds, with a specialized classifier operating in approximately 2 milliseconds. Reduced the complete data-read-to-action pipeline—from reading auction data and calculating bids to deciding and executing actions—from approximately 20 seconds to under 500 milliseconds. Replaced a legacy automated bidding engine, reducing the customer’s operational costs by approximately US$50,000 annually.
Galidor CRM — Multi-Tenant Business Platform
My contribution
Built a full-stack CRM using Node.js, Express, TypeScript, Prisma, PostgreSQL, Vue 3, Pinia, TanStack Query, and Tailwind CSS. Implemented multi-tenancy, tenant-scoped authorization, role-based access control, authentication, webhooks, and workflows for managing customers, leads, projects, and teams. Designed the platform to support integration with external services and provide a foundation for automated customer interactions and agent-driven workflows.
BIA Orchestrator & WhatsApp Automation
My contribution
Developed AI orchestration and conversational workflows using Go, Vue 3, PostgreSQL, and MCP-based tool execution. Designed a webhook- and SSE-based messaging architecture connecting inbound WhatsApp messages, cross-service processing, CRM workflows, agent execution, and response delivery. Replaced the previous WhatsApp delivery microservice with a drop-in replacement integrated with the CRM and agent orchestration architecture. Integrated conversational automation with CRM data to support lead handling, scheduling, sales workflows, and service operations while keeping customer information synchronized.
Results
Reduced end-to-end messaging-loop latency during peak hours from approximately 4 minutes to 10 seconds, improving response throughput and service responsiveness. Delivered approximately US$25,000 in annual operational cost savings through the replacement architecture.
Skuld — Conversational Scheduling Platform
My contribution
Developed a scheduling platform using Go, Gin, GORM, PostgreSQL, Vue 3, and TypeScript, integrating WhatsApp messaging with AI-driven workflows. Implemented agent workflows for lead qualification, appointment booking, rescheduling, and cancellation, with Google Calendar OAuth and synchronization. Integrated messaging, contact, chat, and booking workflows through service-level integrations.
Infrastructure & Architecture Ownership
My contribution
Independently managed application architecture and infrastructure across internal systems and customer deployments. Operated Linux VPS environments and containerized services, including Docker deployments, Nginx reverse proxies, HTTPS/TLS, storage integrations, service networking, and monitoring. Managed and optimized microservices, database connectivity, deployment workflows, and WhatsApp messaging infrastructure. Focused on production reliability, service security, latency reduction, operational efficiency, and infrastructure cost optimization.