Selected Work

The problem

A North American jewelry-retail marketing agency running on tribal knowledge — no formal OKRs, manual ad monitoring, fragile vendor management, rising AWS spend, and no AI in the operating model.

What I built

• Defined company-wide 2026 OKRs/KPIs across 10 functions (Sales, AM, DPM, SEO, Social, Web, Design, Marketing, Ops, Finance).
• Built a Google Maps Places API lead-generation tool end-to-end — vibe-coded with Claude + GCP.
• Shipped agentic AI for real-time Google & Meta Ads monitoring, with daily Slack briefings to the DPM team.
• Shipped AI Voice Agents (Cloudtalk) for first-contact and qualification calls.
• Restructured operations: HubSpot + Jira automations, vendor consolidation, and an AWS cost plan.

The problem

21 jeweler hosting environments plus internal services on AWS at $3,496/month ($41,948/year) — including end-of-life Magento 1 / PHP 5 servers, an orphaned disk, four unattached Elastic IPs, and no Savings Plan.

What I built

• Audited every line item — RDS, EC2, EBS, CloudWatch, NAT, S3 — mapped to client domains.
• Designed a 3-phase plan: Quick Wins (week 1) → Medium effort (1–2 months) → Strategic (next quarter).
• Wrote per-task AWS CLI commands so the dev team could execute without further design work.
• Framed for the CEO across four dimensions: cost, PCI compliance, reliability, and per-client margin.
• Added cost-allocation tagging that unlocks per-client AWS cost reporting.

The problem

Targets existed per function, but there was no integrated company-wide goal framework, no monthly cadence, and no way to see the system performing vs. plan in real time.

What I built

• Designed OKRs across 10 functions, each with a monthly target, quarterly result, % of expected, and YoY growth target.
• Wired up a Marketing KPI scorecard (followers, interactions, email open/CTR, leads, traffic, organic, conversion).
• Tracked weekly sales by client, AM, and sale type — new, upsell, cross-sell, and coop.
• Established a quarterly review cadence with the leadership team to adjust targets and reset commitments.

The problem

Four consumer MVNO brands (Red Pocket, FreedomPop, Karma Mobility, Unreal Mobile) with inconsistent checkout UX, low conversion, a 1.6 App Store rating, and no data warehouse — every dashboard an ad-hoc query.

What I built

• Shipped a unified checkout flow used across all four brands.
• Led an AI chatbot build with the Google Cloud team — delivered the highest-ever company CSAT.
• Built the data warehouse from scratch with IT; integrated BI across BigQuery, Looker, Tableau, and DOMO.
• Led CRM, UX, QA, and Data Science specialists and coordinated the 4-brand design system.
• Shipped ACP enrollment and lost-eligibility flows for FreedomPop.

The problem

iFood Groceries Retailers were on a fragile monolith that couldn't scale to new partner integrations, made fraud signals invisible, and created friction between Product, Engineering, Risk, and Data.

What I built

• Led 2 technical squads + 3 PMs across the Financial, Payments, Retail, and Partners tribes.
• Built the operating model: biweekly sprints, weekly demos, monthly prioritization, quarterly OKR resets.
• Rebuilt the Retail-pickers platform from monolith to microservices.
• Designed an anti-fraud rule rooted in SQL analysis of partner behavior and loss patterns.
• Created the first Partner Analytics Dashboard for Groceries Retailers with the data team.