As a freelance applied AI software developer, I built a payroll-slip extraction pipeline that generated YAML extraction rules from sample documents instead of sending each document to an LLM. It achieved 100% accuracy on a sample of 2,312 slips, with deterministic validation, record-level traceability, and personal-data masking.
I also developed a salary-extraction API using Claude and OpenAI, with confidence-based referral for manual review and no storage of customer data. Its integration contract supported a Java 8 consumer using RabbitMQ retries and a dead-letter queue.
For a WhatsApp customer-service assistant, I orchestrated a five-screen WhatsApp Flows form, lead classification, and automatic monthly reporting. I also developed a serverless insurance module for Vida Ley and SCTR policy issuance and renewals, and a customer onboarding platform with eight stages and operation auditing.
At Smart H.R., I automated digital-certificate expiry notifications and developed a Java report generator from JSON form configurations. I also analyzed platform incidents and supported AWS operations. Earlier, as a freelance programmer, I developed a Python and SQLite inventory application and collaborated on a responsive website.
