BACKGROUND
~ / ignacio.coffee — about:me
pixel coffee beans

ignacio.coffee

Caffeine-based engineering solutions.

SR. FULL-STACK PRODUCT ENGINEER · SOFTWARE ARCHITECT

loc   New York City
from  Chile 🇨🇱
dog   Aki 🐕
stat  ● available
ignacio_aki.jpg
Ignacio with Aki
// ABOUT

Engineer from Chile, currently living in New York City. Big fan of dogs, coffee, pop culture, and video games. My background is in Computer Science, Civil Engineering, and Math.

A Software Architect and Senior Lead Engineer with deep experience building cloud-native, distributed systems. I've architected mission-critical microservices platforms in Node.js/TypeScript, React, Vue, Ruby and Python, deployed on AWS with infrastructure-as-code via Terraform, full CI/CD and containers.

Previously I designed enterprise SaaS platforms serving 900+ organizations, led cross-functional international teams, and consulted on architecture for large-scale government systems.

// STACK
Languages
TypeScriptJavaScriptPythonRubySQLNoSQL
Frontend
ReactVue.jsNext.jsTailwind CSSVite
Backend
Node.jsExpressRuby on RailsSinatraFastAPIRESTGraphQL
Cloud · AWS
ECS FargateLambdaRDSElastiCacheS3CloudFrontSQS / SNS
Infra · IaC
TerraformDockerPodmanKubernetesRenderCloudflare
OS
Arch - Debian LinuxKernel compilationOS Administration
CI/CD
GitHub ActionsJenkinsDockhandDatadogGrafana + Prometheus
Data
PostgreSQL (w/pgvector)SupabaseRedis - ValkeyMariaDBMongoDB - CouchDB
Architecture
MicroservicesEvent-drivenDistributed systemsDomain-driven designDesign Patterns
AI Systems
LLM orchestrationProduction AI systemsRAGVector searchEmbeddingsSemantic searchMCPWorkflow automationML design, training, and prediction
▸ AI SYSTEMS & PIPELINES — CURRENT FOCUS in production

I design the systems and the pipelines.

Most of my work now is production AI: how models retrieve, reason, and act, and the data and orchestration pipelines that make them reliable. I build the tooling, not just the features.

RETRIEVAL PIPELINE
Ingest →Embed →Index →Retrieve →Rerank →Reason →Act
AGENT LOOP
Plan →Act →Observe →Verify →Ship
Hybrid product search
BM25 + vector retrieval with LLM enrichment over a live commerce catalog.
FastAPIMarqoRedisGPT-5-nano
Multimodal vision pipeline
Object detection, classification, and image similarity search, served through an async, queue-backed API.
YOLOGPT-4oGoogle VisionBullMQ
Structured extraction
Business data pulled from PDFs into typed, validated schemas.
GeminiPythonStructured output
Analytics data pipeline
Event ingest → materialized views → SSR dashboards at 100K+ events.
PostgreSQLpg_cronRPCETL
Agentic dev workflows
Multi-agent orchestration, custom skills, and MCP servers that ship real work.
Claude CodeMCPMulti-agent

▸ always on the lookout for the next big challenge.