Juan Valentin Alducin Vázquez
Senior Backend Engineer | Tech Lead · SRE & Automation | Agentic AI
CDMX, Mexico · English B1 · B2 technical reading/writing
Senior Backend Engineer and Tech Lead with 10+ years in Python, SQL and AWS. I help make operations reliable and automated — blameless post-mortems, monthly uptime/downtime tracking, runbooks and n8n automations that cut manual work — and I build software with Spec-Driven Development (SDD/OpenSpec) and AI agents (Claude Code). Previously, technical lead of international GK POS rollouts across Latin America.
How I Build with AI
- Coordinating the Service Support team at Podemos Progresar (fintech): incident resolution with SDD and AI.
- Running my projects locally with Docker; observability via Grafana.
- Pushing my English toward B2 (Quick Learning, in progress).
I work AI-native: Generative AI is the central copilot of my SDLC, not an afterthought. My practice is Spec-Driven Development (SDD/OpenSpec) — the spec is the source of truth, and every change flows through versioned artifacts before a line of code is written.
- Automate > operate. Rule of two: any manual report or task done more than twice becomes a spec-driven pipeline.
- Reliability as a product. Downtime is measured by real user impact, not by ticket closure; blameless post-mortems; least-privilege runbooks.
- State outside the code. IDs, cut-offs and thresholds live in configuration (JSON / Notion) read alike by the workflow and the dashboard — the spec is the contract.
- Cursor over ingestion, not the clock. Designed for failure (missed runs, late data), not for the happy path.
Spec-Driven Development
Requirements and verifiable scenarios drive the work. AI generates specs, analyzes code statically and proposes implementations against the contract — shortening review cycles.
AI as SDLC copilot
Claude (Anthropic), Gemini and OpenAI across the loop: specs, analysis, debugging, RCA & post-mortems and technical documentation — plus an AI Skill in progress to classify and route tickets, with human review when confidence is low.
Operational SLAs & practices
- <10 min first response · <2 h critical resolution
- Blameless post-mortems (P0–P3) with a tracking dashboard
- Monthly uptime/downtime tracking by real user impact
- n8n automations for repetitive operational work
Agentic & machine-readable
Agentic workflows with RAG and AI agents. This very site ships structured JSON-LD and a /llms.txt so agents and LLMs can read my profile directly.
Experience
Service Support Tech Lead & AI Specialist
- Coordinate the Service Support team (bugs, critical incidents, technical debt); applied Spec-Driven Development (SDD) with the Claude API to the diagnose → implement flow to speed up reviews.
- Take part in building the post-mortem process (P0–P3, 5 Whys, blameless), with a tracking dashboard for delays, escalations and progress against target.
- Calculate and track monthly uptime/downtime (fleet × real minutes; downtime = real impact) under the methodology set by leadership; standardize runbooks and least-privilege access.
- Automate part of the operation with n8n: Freshdesk → Notion → Slack, a post-mortem delay watcher and ticket-spike alerts. Use AI (Claude, OpenAI, Gemini) for specs, analysis, debugging and documentation, and I am developing (in progress) an AI Skill to classify and route tickets, with human review when confidence is low.
- Take part in the weekly release trains coordinated by Engineering, validating deployments from Service Support across core banking, origination and the field app. Diagnose and operate AWS (CloudWatch Logs Insights, RDS) and review Sentry, using the diagnostic and post-deployment skills built by the infrastructure lead to catch post-deploy errors and estimate impact. SLA targets: first response under 10 min, critical resolution under 2 h.
- Stack: Python · SQL · PostgreSQL · AWS (CloudWatch Logs Insights, RDS) · Sentry · n8n · Claude API · Freshdesk / Notion / Slack
Backend Support Specialist
- Led N2 incident management, producing RCAs and AI-assisted technical post-mortems that prevented recurring failures; enabled autonomous resolution in the N1 team through knowledge transfer.
- Own the weekly idempotent cleanup ETL on PostgreSQL (backup → delete → update → audit log, pre-validated against expected counts) and identity deduplication in production; schema changes versioned with Liquibase.
- Central technical liaison in operational coordination, accelerating incident resolution across Python, Django, FastAPI and PostgreSQL systems.
- Stack: Python · SQL · PostgreSQL · MySQL · Liquibase · Django · FastAPI · AWS · Docker · Grafana
Software Engineer → Tech Lead
- Developed REST APIs (Python, Java, PHP) integrating GK POS with ERPs/CRMs through XML/SFTP automation for retail across 3 LATAM countries, adapting solutions to distinct cultural and regulatory contexts.
- Technical lead for the cloud GK POS Go-Live (Peru, Colombia, Bolivia): coordinated timelines, managed business and IT stakeholders, ran technical pre-sales demos and trained local teams.
- Designed the promotions ETL pipeline for GK OmniPOS, enabling real-time commercial analysis for decision-making.
- Led the functional-technical GK POS team: defined the testing strategy (unit, regression, integration) and ensured end-to-end quality on every release.
- Stack: Python · Java · PHP · GK POS · GK OmniPOS · XML · SFTP · Docker · ETL
Software Engineer
- Implemented billing systems, customer portals and ETLs integrated with SAP ERP/HANA; built executive dashboards in SAP BO communicating key metrics to business leadership.
- Provided AMS support to Retail GK solutions: help desk, corrective and evolutive maintenance of applications and ETL flows.
- Collaborated in cross-functional teams under Scrum, maintaining a 98% SLA.
- Stack: Python · PHP · .NET · SAP ERP/HANA · SAP BO · ETL · Web Services · Scrum
★Featured Projects
★ 100% built with Spec-Driven Development (SDD/OpenSpec) — the spec is the source of truth in every project.
platform-STALD — Personal Project
An educational platform I built as a hobby and use to practice technologies and AI: Spec-Driven Development with Claude Code (subagents, skills and hooks), real-time multiplayer games, Supabase Auth and Postgres with RLS, and CI tests.
Fidello — Loyalty Card System
Multi-business digital loyalty (mobile-first QR, nothing to install) with all the business rules in PostgreSQL: multi-tenant RLS, PL/pgSQL RPC, anti-replay QR tokens and feature flags. 100% SDD/OpenSpec with 5 specialized AI agents.
Monitoreo-Cloud — Observability Stack
Self-hosted observability stack (Grafana + Docker) with a single pane over three sources: AWS serverless (Lambda invocations/errors/duration + S3) via CloudWatch, local Docker containers (CPU/memory/health), and Trackion's tickets & SLA. Automated collection with n8n + a read-only IAM role. 100% SDD/OpenSpec.
VoltGrid — EV Charging Platform
Multi-tenant, white-label SaaS for EV charging-station operators: live station status over WebSocket (polling fallback), tenant isolation via PostgreSQL Row-Level Security (RLS), RBAC (superadmin/org_admin/operator/viewer) + SSO (OIDC), analytics (uptime, kWh, CSV export), an auditable state scheduler and an installable PWA. 100% SDD/OpenSpec.
Trackion — Serverless Help Desk
White-label serverless help desk: end-to-end ticketing (categorize, prioritize, assign, history; open→in_progress→resolved→closed), admin-managed catalogs, and an open API-integration module (generic inbound webhooks + outbound domain-event dispatch) to plug any external system without touching the core. Grafana tickets & SLA dashboards. 100% SDD/OpenSpec; runs locally via Docker.
dataMasterGK — Retail ETL Middleware
ETL middleware that automates master-data integration into GK Software retail systems: watches directories, transforms Excel → GK XML across four interfaces (stores, operators, promotions) and transmits via SFTP/FTP — with a Flask web panel (config, operation, audit), a scheduler and SQLite auditing. Re-engineered under SDD/OpenSpec: idempotent single-pass pipeline, pooled SFTP/SQLite connections, and security hardening (parameterized/whitelisted SQL, secrets out of the repo).
✎Writing
Notes from building AI-native: Spec-Driven Development, Docker and observability — drawn from real projects. Project deep dive: Fidello.
Anatomy of a change: one real SDD/OpenSpec change, artifact by artifact
The full flow (proposal → specs → design → tasks → apply → verify → archive) on a real change of this repository: 8 specs, 8 design decisions, 47/47 automated checks and a CV calibrated to exactly one page.
Read ↗Spec-Driven Development: the spec is the source of truth
Why I treat the spec as the contract and code as its current implementation — the artifact flow (proposal → specs → design → tasks → apply → archive), and why it shortened my review cycles.
Read ↗A zero-cost local observability stack with Docker Compose
From an always-on EC2 to a reproducible local stack (Grafana + PostgreSQL + n8n): $0 idle, idempotent, and still cloud-aware over a read-only IAM role.
Read ↗Multi-source observability with Grafana
One pane over three sources: AWS serverless (CloudWatch), local Docker containers, and application data (tickets & SLA) — correlating infra and business signals.
Read ↗Hard Skills
Tech Stack by level of mastery
An honest inventory, not a keyword list. Own = built end-to-end and operationally owned · Build = solutions I assemble recurrently · Operate = fluent for diagnosis and operations, not building the platform · Learning = directed study, no operational depth yet.
Own
Built it end-to-end; if it breaks, they call me.
- PostgreSQL & advanced SQL
- Idempotent production ETL
- Runbooks & least-privilege access
- Notion as knowledge system
- Spec-Driven Development (OpenSpec)
Build
I assemble solutions with these on a recurring basis.
- Python
- FastAPI
- Django
- Flask
- pytest (mocks)
- SQL
- Liquibase
- n8n (workflow SDK)
- Freshdesk / Notion / Asana / Slack APIs
- Webhooks · serverless tooling
- Spec-driven pipelines
- Grafana dashboards & alerting
- Chart.js dashboards
- Claude API
- Claude Code (subagents, skills, hooks, MCP)
- OpenAI API
- Gemini API
- Amazon Bedrock
- Prompt Engineering
- LLM Integration
- AI Agents / Skills
- Agentic Workflows
- RAG
- SonarQube
- GitHub Actions (CI/CD)
- Docker
- Serverless Framework
- Supabase (PL/pgSQL, RLS, Auth, Realtime, Edge Functions)
- Deno / Deno Deploy
- React
- TypeScript
- Next.js
- Node.js
- Vitest / Jest
- SQLAlchemy 2.0 async · Alembic · Pydantic v2
- Pandas · openpyxl · SFTP / Paramiko
- OpenAPI · REST · WebSocket · SSO / OIDC
Operate
Fluent for diagnosis and operations; I don't build the platform.
- AWS Lambda
- Step Functions
- CloudWatch Logs Insights
- RDS
- S3
- EventBridge
- EC2
- AWS SSO / Identity Center
- Sentry
- Post-mortem process (P0–P3, blameless)
- Monthly uptime/downtime tracking
- MySQL
- SQL Server
- MongoDB
- DynamoDB
- SAP ERP / HANA · SAP BO · SAP CAR
- GK POS · GK OmniPOS
- Core banking / credit-bureau integrations
- Scrum · Jira · Asana · Postman
- Azure (base)
- GCP (base)
Learning
Directed study; no operational depth yet.
- IaC: Terraform / AWS CDK
- Amazon QuickSight
- Synthetic checks & proactive observability
- AWS Glue at scale
Soft Skills
- Leadership
- Effective Communication
- Teamwork
- Requirements Analysis
- Analytical Thinking
- Adaptability
- Problem Solving
- AI-native Workflow
Languages
| Spanish | Native |
| English | B1 — Intermediate (reading, writing & technical communication) |
Education & Certifications
Computer Systems Engineering
Instituto Tecnológico de Orizaba · 2006 – 2011
Orizaba, Veracruz, Mexico
Advanced English
Quick Learning · In progress
Online
Certifications & Courses
- Claude Code in Action · Anthropic · 2026
- Claude Code 101 · Anthropic · 2026
- Introduction to Model Context Protocol · Anthropic · 2026
- Introduction to Agent Skills · Anthropic · 2026
- Introduction to Subagents · Anthropic · 2026
- Claude Code: Software Engineering with Generative AI Agents · Vanderbilt University · 2026
- Generative AI Applications with Amazon Bedrock · AWS · 2026
- Developing Applications in Python on AWS · Coursera · 2025
- GitHub Actions Bootcamp · Código Facilito · 2025
- Understanding the World Through Data · edX · 2024
Contact
Looking for job opportunities? Let's talk!
Or send me a message:
Terminal
A real mini-terminal. Type help and press Enter. 😉