End-to-end B2B SaaS platform that automates public procurement monitoring (SECOP II) in Colombia. Ingests and analyzes thousands of government bidding opportunities, delivering real-time email alerts and analytics to paying enterprise clients.

Global emergency intelligence engine combining high-throughput Go ingestion pipelines with Python semantic retrieval. Uses PostgreSQL + pgvector (HNSW indexing) for seismic hazard queries, validated through a DeepSeek AI gatekeeper.

Autonomous software engineering skill that orchestrates multi-agent workflows across 7 specialized phases—from architectural design and implementation to automated verification, strict invariants, and clean delivery.

Production engineering knowledge base and playbook covering concurrency invariants, streaming patterns, context propagation, container CPU/CFS limits, and backend reliability best practices.

Clean-architecture Go microservice platform for AI-powered card-based book distillations. Transforms technical PDFs into structured cards while strictly preserving author voice, verbatim quotes, and original diagrams.

Distributed, strongly-consistent feature flagging system powered by the Raft consensus algorithm. Supports dynamic leader election, log replication, and replicated state machines to eliminate single points of failure.

High-throughput append-only storage engine built in Go for time-series workloads. Implements an append-only Write-Ahead Log (WAL), in-memory SkipList memtable, SSTables with sparse index and Bloom filters, and background compaction.

High-scale Go microservices and event pipelines handling 30K+ daily state updates for Disney media platforms. Optimized throughput by 2× while reducing infrastructure costs using AWS Kinesis, EventBridge, and MongoDB.

Scalable event-driven Go backends processing tens of thousands of daily events with Google Cloud Pub/Sub, ensuring high availability, zero message loss, and horizontal elasticity across Latin America.

Hybrid recommendation system combining collaborative filtering with content-based metadata to solve the cold-start problem for new users, delivering accurate recommendations even without interaction history.

Automated customer support routing engine that analyzes incoming tickets with NLP classification, scoring sentiment and urgency to route high-priority issues to the appropriate engineering teams.