Projects

Problem

Legacy ERP systems couldn't communicate in real-time, forcing nightly batch jobs and causing stale data across departments.

Solution

Built an event-driven integration platform using Spring Boot and Apache Kafka that processes over 50k transactions/day with sub-200ms latency and 99.9% uptime.

Problem

A single-node cron system was becoming a bottleneck, failing silently and providing zero visibility into job status.

Solution

Designed a distributed scheduler with leader election (ZooKeeper), dead-letter queues, retry policies, and a React dashboard for real-time monitoring.

Problem

Existing tools had no good support for B3 assets alongside international portfolios, forcing manual spreadsheet updates.

Solution

Built a full-stack app integrating B3 data feeds, currency conversion APIs, and a React dashboard with charts for portfolio allocation and dividend history.

Problem

A growing startup was burning $30k/month on AWS with no visibility into waste.

Solution

Built a tool that analyzes AWS Cost Explorer data, tags underutilized resources, and generates actionable reports delivered weekly via Slack.

Problem

Notification delivery was unreliable and tightly coupled to the monolith, causing cascading failures during peak traffic.

Solution

Extracted notifications into a dedicated microservice with priority queues, rate limiting, and fallback channels — achieving 99.95% delivery rate.

Problem

Each team had hand-crafted pipelines — inconsistent, hard to maintain, and no security scanning.

Solution

Built a composable GitHub Actions library with shared workflows for Java services: lint, test, SAST, Docker build, and Kubernetes deploy in under 10 minutes.