Luis Angel Carbonel Hidalgo

Software Engineering Consultant · Austin, TX

• Built and launched BriteCover (britecover.com), a multi-tenant SaaS CRM now in production at insurance agencies—replacing spreadsheet-based lead and renewal tracking with 9 AI features including 0–100 lead scoring, renewal churn-risk analysis, cross-sell detection, and an agentic assistant with 12+ tools that queries the CRM conversationally.
• Built and launched SmartFarmPilot (smartfarmpilot.com), a multi-tenant farm management SaaS serving 100+ users, replacing paper and spreadsheet record-keeping with FEFO inventory allocation, livestock health tracking, CSA subscription management, and AI-powered insights on AWS Bedrock.
• Designed an AI-powered post-sale support architecture for a high-volume e-commerce operation (~50K orders/month) and built an agentic conversational analytics system—intent classification via OpenAI, specialized action routing, and Odoo ERP API integration—replacing manual reporting for the management team.
• Ran each product end to end: architecture, implementation, AWS deployment, and the monitoring that keeps them running.

• Next.js 16
• React 19
• Prisma
• PostgreSQL
• Vercel AI SDK
• AWS Bedrock
• OpenAI
• Stripe
• Resend
• Cloudflare Workers
• Strands Agents
• Odoo

Full Stack Software Engineer (Consultancy) · Austin, TX

• Architected and deployed a production RAG system processing 1M+ queries/day at 99.9% uptime, reducing AI hallucinations by 75% through hybrid vector retrieval (pgvector) and custom relevance scoring.
• Built the vector embedding pipeline behind it, ingesting 100K+ documents/hour from the Redshift warehouse, plus an observability stack tracking RAG quality metrics in production.
• Designed and delivered a multi-provider AI Agent Composition platform letting non-engineers assemble an orchestrator, attach subagents up to 5 levels deep, and bind tools and domain knowledge bases per agent—all at runtime, with no code redeploys—now used by the marketing team and company board for self-serve analysis and by data engineers to validate data and queries.
• Built a Natural Language-to-SQL module (QueryGPT) bridging LLMs with Redshift data warehouse, eliminating manual query writing for data analysts across the organization.
• Built the full-stack Composer UI (admin studio + chat interface) with Next.js 16, React 19, and TanStack Query: provider-aware model selection, session routing, and custom streaming transport compatible with Vercel AI SDK v6.
• Engineered a dynamic agent orchestration engine with agents-as-tools delegation, dynamic tool loading from a registry, and hierarchical agent composition with cycle detection and depth-limit enforcement for safe recursive execution.
• Implemented real-time streaming execution with tool-call tracking, multi-turn conversational memory (sliding-window context), and secure credential management via AWS SSM Parameter Store—ensuring zero API key exposure across the platform.

• Next.js 16
• React 19
• Python
• FastAPI
• Strands Agents
• PostgreSQL
• pgvector
• Java
• Spring Boot
• OpenAI
• AWS Bedrock
• Redshift
• Airflow

Senior Software Engineer III · Montevideo, Uruguay (Remote)

• Replaced the marketing team's manual bid generation with a fully automated bid optimization pipeline, leading a 3-person team (2 engineers + 1 mathematician) across Despegar's meta-search strategy—Trivago, Google Ads, Kayak and other travel platforms.
• Extracted the per-vertical bid formula out of the monolithic ETL job into a dedicated Kotlin/Vert.x microservice with Protobuf serialization, isolating the IO-bound external API calls so each bidding platform could scale independently of the batch run.
• Built real-time monitoring dashboards with Grafana and Loki, including anomaly detection with automated email alerts enabling the team to pause/resume ETL workflows at any failure point.
• Conducted technical interviews for engineering hires in the Montevideo office and ran code reviews to keep quality standards consistent across the team.
• Managed the team's backoffice microservice and led the migration of internal web interfaces to React, improving developer productivity and user experience.

• Java
• Kotlin
• Vert.x
• Protobuf
• Spring Boot
• Hadoop
• Spark
• Polars
• Grafana
• Loki
• React
• Jenkins
• GitHub Actions

Software Engineer – Backend · Montevideo, Uruguay

• Built and maintained the merchant integration platform processing ~50K requests/day for restaurant order management across Latin America (13+ countries).
• Developed and maintained multi-language SDKs (Java, Kotlin, JavaScript, Groovy) used by thousands of restaurants for order ACKs, menus, and real-time notifications via AWS SNS/SQS.
• Built reactive microservices with Ktor and Kotlin coroutines for high-throughput, non-blocking request processing, ensuring low-latency order acknowledgment and delivery status updates.

• Kotlin
• Ktor
• AWS SNS/SQS
• Docker
• New Relic
• Graylog

Software Developer Engineer · Montevideo, Uruguay

• Led migration of transaction processing—20–30K transactions/day—from a Java EE monolith to Spring Boot microservices with Kafka event streaming, Eureka service discovery, and centralized config via Spring Cloud Config.
• Implemented secrets management with HashiCorp Vault, full observability stack with ElasticSearch, Kibana, Grafana, and Prometheus, and established development standards and best practices for the microservices migration.
• Built a real-time transaction monitoring dashboard (JSF/PrimeFaces) tracking EJB container metrics and merchant processing KPIs, improving incident detection and response times.

• Java
• Spring Boot
• Spring Cloud
• Kafka
• Eureka
• HashiCorp Vault
• ElasticSearch
• Kibana
• Grafana
• Docker