leadership
[4]Systems for teams, products and delivery.
- portfolio-strategy 7 self-service capabilities
- team-systems 3 teams · ~30 people
- cross-team-delivery regional dependencies
- product-platform-thinking journeys + foundations
company : "Falabella"
domain : "Customer & Seller Self Service"
scope : "7 capabilities · 3 teams · ~30 people"
practice : "Harness Engineering"
# Core identity configuration
name = "Lucas Andre Henry Diaz"
role = "Tech Team Lead"
company = "Falabella"
location = "Santiago, Chile"
# Specialization
focus = "Customer & Seller Self Service"
scope = "7 capabilities / 3 teams / ~30 people"
# Languages
lang_primary = "es_CL.UTF-8"
lang_secondary = "en_US.UTF-8" # C1 Advanced
# AI Integration
harness_engineering = true
ai_tools = ["Herdr", "Claude Code", "Codex", "Oh My Pi", "Vercel AI SDK"]
"Build systems where people and agents can ship with evidence."
No percentages or vanity versions—just the systems I lead, build and operate.
5 modules / 21 capabilities Systems for teams, products and delivery.
The durable system around replaceable agents.
Different runtimes for different jobs.
Full-stack systems for real commerce journeys.
Evidence before claims.
From hands-on commerce engineering to leading the full self-service portfolio.
self-service (lead/self-service) HEAD @ Falabella
Leading technology strategy and delivery across the customer and seller self-service portfolio in South America.
+Seven capabilities: Returns, Cancellations, My Purchases, My Cases, My Account, Seller Returns and Seller Support
+Three engineering teams and roughly 30 people across shared customer journeys
+Agent-ready repositories, automated quality gates and observable end-to-end flows
+Cross-team alignment across identity, orders, payments, support and mobile dependencies
+Building an internal agent cowork with Vercel AI SDK and human verification
post-purchase (lead/post-purchase) @ Falabella
Led the architecture and engineering practices behind regional post-purchase journeys and platform modernization.
+Microfrontends architecture and monorepo adoption across multiple businesses
+Returns, cancellations and order-management customer journeys
+Playwright E2E framework and automated delivery practices
+Product health and observability with Grafana
falabella-senior (engineering/platform) @ Falabella
Improved the digital commerce platform with backend optimizations and Go services for cross-system communication.
+My Purchases experience and performance improvements
+Go microservices and gRPC contracts
+Next.js and React monorepo delivery
+A/B testing, GitLab CI and Datadog
cencosud-senior (marketplace/logistics) @ Cencosud
Built Seller Center modules for Paris Marketplace and designed an API-driven transport management system.
+Seller Center modules for marketplace operations
+Transport Management System and multi-carrier logistics
+Complex business-rule automation
cencosud-engineer (marketplace/orders) @ Cencosud
Focused on asynchronous order management and legacy commerce integration.
+Asynchronous seller-order management
+XML processing automation
+Legacy platform integration
witi (commerce/returns) @ WiTi
Integrated PIM systems and led returns-module development using explicit state machines.
+PIM integration for Paris
+Returns workflow with state machines
+Marketplace feature delivery
# falabella-microfrontends
Microfrontends architecture supporting multiple businesses across South America with monorepo system.
## Stack
# playwright-e2e-framework
E2E testing framework with automated regression tests and market-specific test generation.
## Stack
# analytics-engine
Proprietary analytics framework capturing frontend events without third-party dependencies.
## Stack
# transport-management-system
API-driven TMS optimizing multi-carrier logistics with automated carrier selection.
## Stack
Agents are replaceable. The durable advantage is the system around them: memory, orchestration, scoped tools, observability and human judgment.
A durable loop for thinking, delegating and verifying across personal work.
A shared workspace for agents and engineers, designed around real team workflows.
Decisions and sources live outside any single model session, ready to be reused and challenged.
Herdr coordinates specialized agents; workflows define boundaries, handoffs and recovery paths.
Tests, evals, traces and human approval turn an agent's claim into an accountable deliverable.
echo $OPERATING_PRINCIPLE“I don't optimize for more AI output. I build the harness that makes useful output trustworthy.”
lucas @ portfolio : ~ $ cat ~/.ssh/available_ports
Let's build a system that can earn trust together.