6
projects in production
Senior .NET & Microservices Developer
Héctor Sulbarán
AI-Augmented Engineer
I build enterprise backend systems, scalable APIs, and AI-agent workflows that help teams ship faster with technical precision.
8+ years. 4 countries. Real production environments.
Methodology
Human-led engineering. AI-augmented execution.
AI accelerates delivery inside a system of context, constraints, and verification. Architecture, security, product decisions, and final acceptance remain under human judgment.
< 24h
minimum time to production (Faro)
99.7%
latency reduction (HorasBot)
3×
commit velocity (NetSuite)
Keep critical paths independent
External dependencies and supporting layers stay decoupled so the core function continues operating when an integration fails or becomes unavailable.
Meet users in channels they already use
Telegram, WhatsApp, or Slack reduce friction when they already belong to the user's workflow. Architecture adapts to real behavior instead of forcing a new interface.
Use AI only where it adds value
Deterministic tasks remain deterministic. AI reasoning is reserved for ambiguity, analysis, and complex execution; it is never added by trend or by default.
Validate before assuming
Assets, data, requirements, and system responses are verified against real behavior. No input is treated as complete simply because it appears correct.
Require human review at every phase
Planning, implementation, and delivery advance in stages with explicit approval. Every phase has acceptance criteria, technical verification, and a clear point to stop or correct course.
Treat living documentation as the source of truth
Briefs, specs, plans, acceptance criteria, and CLAUDE.md hold context, decisions, and constraints. Execution starts from that documentation system, never from an isolated prompt.
Design governance that prevents errors
Scoped permissions, traceability, guardrails, and explicit approvals turn technical judgment into operating rules. The system does not depend on the model remembering to behave correctly.
Cross-Project Summary
One method applied across six different contexts.
| Project | Type | Delivery pressure | Headline metric |
|---|---|---|---|
| Faro | Humanitarian response | Extreme · < 24h to production | 500+ · 89% · < 3s |
| HorasBot | Internal enterprise automation | Medium · daily team adoption | 15 min → 3 s · 99.7% · 98% |
| Ratio | Personal finance product | No external pressure · full SDD | 0 · 120+ · 6 |
| WOI 2469 | B2B/B2C commercial landing page | Medium · client delivery cycle | 4 B2B · 3d · 10× · ≥95 |
| hsulbaran.dev | Personal brand and lead conversion | Low · detailed visual iteration | Lighthouse ≥90 · $0/mes · bilingüe · < 3 semanas |
| NetSuite | Enterprise AI governance | Sustained · active production system | 482 · 125 · 1–8 → 18–23 |