Projects using AWS in San IsidroProjects using AWS in San Isidro
Projects
People
Cover image for Cutting a 29-Product Portfolio Down
Cutting a 29-Product Portfolio Down to What Actually Sells Problem. A portfolio of 29 products grown organically from client requests. An audit found that none of them were commercially ready to sell: no pricing model, no sales enablement, almost no documentation. Engineering capacity was spread across products nobody could monetize. Action. Ran a full portfolio triage against strategic value and commercial viability, landing on four dispositions: keep, evolve, research, kill. The hard part was not the analysis — it was that several kill candidates still carried live revenue, so each one needed a migration path rather than a shutdown date. Paired the triage with an operating model change: no new product enters a build sprint without a named first client and a signed-off commercial definition. Result. Portfolio reduced to a focused set with explicit migration paths for everything retired, and a governance gate that stopped the pattern from repeating.
1
203
Cover image for From 7 Fragmented Data Sources
From 7 Fragmented Data Sources to One Customer Identity Model Problem. Seven disconnected Data Sources, no single customer identity, no traceability. Every business question turned into a reconciliation exercise before it could become a decision. Action. Designed a Single Source of Truth with one customer identity model and explicit data stewardship. Ran the stack decision myself: evaluated AWS S3 + Athena against Redshift and BigQuery on three criteria — total cost at our data volume, existing team skillset, and time to first useful query. Chose S3 + Athena; the cost profile and the zero-ramp for the team beat the query performance advantage of the alternatives at our scale. Built the analytics layer on Power BI + Microsoft Fabric. Result. An LLM-driven BI product adopted by 10 enterprise operators at daily and weekly cadence — the measure that mattered was recurring use, not dashboards delivered.
1
125
Cover image for AI-Powered Dynamic Pricing for a
AI-Powered Dynamic Pricing for a B2B Transportation Platform Problem. Passenger transport operators on a LatAm B2B SaaS platform were setting fares by intuition. No visibility into yield, load factor, or route-level profitability — and no way to act on a pricing signal fast enough for it to matter. Action. Defined and shipped an AI-assisted revenue management system on top of ~20 years of transactional data. Built the pricing signal layer (yield, load factor, average fare, route profitability), then closed the loop so recommendations became executable price changes rather than reports nobody acted on. Prioritized execution latency over model sophistication — the constraint was operator workflow, not prediction accuracy. Result. System in production with 10 enterprise operators, running pricing decisions on data instead of instinct.
1
112