IÑAKI
ECHABURU DUTREN
Software Developer specializing in distributed systems, event-driven architecture, and integrating AI into production-grade applications.
CONTACT ME →CASE STUDIES / DEEP DIVE
Real-Time Event Pipeline
E-commerce platform struggled with processing 500K+ daily orders, causing checkout delays and inventory sync issues.
Design and implement a scalable event-driven architecture to handle order processing, inventory updates, and third-party integrations.
Built distributed event pipeline using Kafka, NestJS microservices, and Redis for caching. Implemented CQRS pattern with PostgreSQL for writes and read replicas for queries.
99.9% uptime, reduced p95 latency from 2.3s to 180ms, processed 2M+ events/day with zero data loss. Saved $40K/month in infrastructure costs.
AI-Powered Data Pipeline
SaaS company needed to extract structured data from 100K+ unstructured customer support tickets monthly.
Build an automated data extraction pipeline using LLMs to classify, summarize, and route support tickets.
Designed workflow orchestration with Temporal, integrated Claude API for entity extraction, and built vector search with Pinecone for semantic ticket matching. Implemented streaming responses for real-time UI updates.
Reduced manual triage time by 75%, improved ticket routing accuracy to 94%, and processed 3.2M tokens daily at $0.003/ticket cost.
RAG Knowledge System
Internal documentation across 12 repos was fragmented, making onboarding new engineers a 3-month process.
Create a conversational AI assistant that provides accurate, context-aware answers from the entire engineering knowledge base.
Built RAG system using LangChain, embedded 50K+ documentation pages into ChromaDB, implemented re-ranking with Cohere, and added citation tracking. Deployed with streaming responses and feedback loops.
Reduced avg. onboarding time to 3 weeks, 89% answer accuracy rate, 2,500+ queries/month, and 94% developer satisfaction score.
Microservices Orchestration
Monolithic payment system couldn't scale during Black Friday, leading to $200K in lost revenue and customer churn.
Decompose the monolith into microservices while maintaining zero-downtime migration and data consistency.
Applied strangler fig pattern, extracted payment, fraud detection, and notification services. Implemented saga pattern for distributed transactions, used Istio for service mesh, and built comprehensive observability with Datadog.
Handled 10x Black Friday traffic (150K concurrent users), reduced deployment time from 2hrs to 15min, and achieved 99.95% uptime SLA.
TECH STACK / 2026
BACKEND
- ■Node.js
- ■Python
- ■NestJS
- ■FastAPI
- ■Express
FRONTEND
- ■React
- ■Next.js
- ■TypeScript
- ■Tailwind
- ■Redux
- ■Zod
DEVOPS
- ■Docker
- ■S3
- ■Linux
- ■CI/CD
- ■Grafana
AI/ML
- ■LangChain
- ■OpenAI
- ■RAG Systems
ARCHITECTURE
- ■Microservices
- ■Event-Driven
- ■CQRS
- ■REST
DATABASES
- ■PostgreSQL
- ■MySQL
- ■Redis
THE HUMAN & AI
ENGINEERING PHILOSOPHY
I build systems that don't just work—they scale, adapt, and survive production chaos. My approach prioritizes observability-first architecture, domain-driven design, and treating infrastructure as a first-class concern.
Every line of code is a trade-off. I document those decisions, build with testing in mind, and believe that the best code is code you don't have to write.
AI COLLABORATION IN 2026
I use AI as a force multiplier—not a replacement. Claude assists with boilerplate, helps explore edge cases, and acts as a rubber duck for architecture decisions. GitHub Copilot accelerates the mundane.
But the hard problems—system design, performance optimization, managing distributed state—those still require human intuition, battle scars, and knowing when to ignore the AI's suggestions.