Projects completed
Product engineer
Spaces published
Active logistics records
Industrial formulas managed
The decisions behind the systems.
These projects show how I move from an ambiguous operating problem to a product people can use and trust.
Founder product · Out-of-home advertising
Publicate
A marketplace and operating platform connecting advertising discovery with the work required to deliver a campaign.
Status
2025Present
Advertising inventory, conversations, proposals, contracts, evidence, invoices, and payments lived across disconnected moments and tools.
Discovery was only the beginning. A useful marketplace also had to preserve context as each opportunity became a proposal, contract, campaign, and financial operation.
One connected product for space owners, brands, and teams, built around a shared domain model from publication through payment.
Evidence
- 011,500+
advertising spaces
- 022026
In production since February
- 03
Built from scratch by two programmers
As co-founder, I share end-to-end product and engineering ownership with one other programmer across architecture, implementation, and delivery.
- 01
Role-aware marketplace and operating workflows
- 02
Contracts, evidence, invoicing, payments, and payouts
- 03
Background jobs, real-time communication, monitoring, and tests
- Next.js
- React
- Node.js
- Prisma
- PostgreSQL
- Redis
Confidential system · Logistics and transportation
Multi-city logistics operations
A shared operational workspace for teams continuously coordinating import, national delivery, and return workflows.
Status
April 2026Production
Operational spreadsheets changed throughout the day. Teams compared files and chat screenshots to understand the latest status, while the reason behind a changed row was easy to lose.
The operation did not need another place to capture rows. It needed one trustworthy operational state that different roles could update safely throughout the day.
Persistent delivery records with controlled editing, automatic lifecycle transitions, shared visibility, conflict protection, and complete change history.
Evidence
- 01100+
active operational records
- 0220–30
internal users
- 03
Four cities · Three connected workflows
As technical co-founder at Vanto Nuvem, I led discovery with the client, designed the operating model, and implemented the production system.
- 01
Permissions by role, city, and field
- 02
Daily carryover, automatic refresh, and concurrent-edit protection
- 03
Field-level audit history and reversible lifecycle transitions
- Next.js
- TypeScript
- Prisma
- PostgreSQL
- Supabase
- Excel
Confidential system · Industrial manufacturing
Industrial costing and monthly close
A traceable costing workflow connecting SAP purchases, materials, formulas, forecasts, decisions, and frozen monthly history.
Status
July 2026Production
Purchase files, formulas, unit conversions, manual costs, forecasts, and monthly history were distributed across spreadsheets, making past decisions difficult to reconstruct.
The company did not only need faster arithmetic. It needed every cost to remain explainable and every closed month to remain historically true.
A controlled monthly process that validates inputs, recalculates dependencies, blocks unsafe closes, and freezes the evidence behind each final cost.
Evidence
- 01150+
industrial formulas
- 02
SAP MB51 ingestion
- 03
Twelve-month forecasts · Frozen monthly snapshots
I translated the costing process into a product model, designed its safeguards, and independently implemented the application through Vanto Nuvem.
- 01
Versioned and nested formulas with cycle detection
- 02
Protected manual costs and explicit unit conversions
- 03
Close blockers, audit history, and frozen period snapshots
- Next.js
- TypeScript
- Prisma
- PostgreSQL
- SAP exports
Personal product · iOS personal finance
Speakance
A voice-first expense tracker designed to make capture and statement reconciliation faster without sacrificing trust.
Status
2026Active product development
Recording an everyday expense feels like form entry, and reconciling a statement later requires repetitive searching, matching, and classification.
Speed alone was not enough. Voice and AI only become useful when uncertain results are reviewable, offline work is preserved, and retries cannot create duplicates.
A native iOS product combining voice, text, and manual capture with reliable synchronization, structured review, insights, and statement reconciliation.
Evidence
- 01
Native SwiftUI application
- 02
Offline-first capture queue
- 03
Local and AI-assisted statement reconciliation
I conceived the product from a problem in my own routine and independently own its product design, iOS engineering, backend, and supporting web experience.
- 01
Idempotent offline queue with ordered retries
- 02
Review states for uncertain AI output
- 03
On-device statement parsing with a redacted, usage-limited AI fallback
- SwiftUI
- Supabase
- OpenAI
- PostgreSQL
- Next.js
More evidence of the same practice.
Two additional Vanto Nuvem systems show how the same discovery-to-delivery approach adapts to commercial data and public-service operations.
Commercial intelligence and revenue
Sales and finance repeatedly reconciled quotations, orders, billing, collections, receivables, and manual classifications from separate ERP exports.
I designed one controlled commercial model with validated imports, explicit financial overrides, recoverable loads, funnel automation, dashboards, and reporting.
- 330 clients and prospects
- 757 quotations · 1,505 line items
- 120 orders plus financial records
Youth program operations
Participants repeatedly submitted the same information while administrators reconciled applications, documents, eligibility, capacity, and follow-up across separate tools.
I centered the product on reusable participant identities, then added program-specific requirements, document controls, approvals, capacity rules, waitlists, and exports.
- Reusable participant profiles
- Private document workflows
- Capacity-aware registration and waitlists
One continuous line of ownership.
I stay close enough to the operation to find the real problem, then remain responsible for the product until it is useful in production.
- 01
Observe
Understand the people, files, exceptions, and decisions that keep the operation moving.
- 02
Diagnose
Separate the visible request from the underlying bottleneck worth solving.
- 03
Model
Turn responsibilities, rules, history, and edge cases into a clear product model.
- 04
Design
Define the smallest dependable release and the workflow people will actually adopt.
- 05
Build
Own the architecture, data, interface, integrations, testing, and deployment.
- 06
Improve
Learn from real usage and evolve the product without losing operational trust.
Problems I chose for myself.
Outside client work, I keep building products around everyday friction, interaction design, and systems I want to understand more deeply.
B1Native iOS · Bowling operations
BowlUp
League scorecards and player trends were difficult to preserve and analyze from physical or photographed boards.
One tested scoring engine governs capture, validation, totals, records, and analytics in a local-first product.SwiftUI · Swift 6 · Local persistence · OpenAI
R2Web product · Music ranking
Rankify
Ordering an entire album or playlist at once creates too many simultaneous and inconsistent decisions.
Pairwise comparisons reduce cognitive load, while an insertion algorithm constructs and exports the final ranking.Next.js · TypeScript · Supabase · Spotify API
Profile
Close to the operation. Responsible for the system.
I studied software engineering at Tecnológico de Monterrey and began in data engineering. Working close to operators pulled me toward the complete product: what to build, how the rules fit together, and what makes a system dependable after launch.
At Vanto Nuvem, my co-founder leads the commercial and legal side while I lead technical discovery, product definition, architecture, and engineering. At Publicate, I share product and engineering ownership with one other programmer inside our four-founder team.
From data foundations to complete products.
Co-founder · Technical lead
Vanto Nuvem
Lead discovery, solution design, architecture, implementation, and production delivery for tailored operational products.
Co-founder · Product engineer
Publicate
Share end-to-end product and engineering ownership of a marketplace and operating platform built from scratch.
Data engineer
Slalom Build
Built multi-source data-platform workflows with Spark, Delta Lake, Databricks, and automated data-quality validation.
Data engineer
Danu Analítica
Built ETL pipelines and client-facing data applications across AWS, GCP, Retool, and business-intelligence tools.
Where does your operation lose time, context, or control?
Bring me the operation before the solution is obvious. No polished brief is required—start with what people do today, what breaks, and what a better system would make possible.
Email Andres ↗