• Built 30+ dashboards in AWS QuickSight and Apache Superset spanning
leadership reporting, 5 departments (Sales, Finance, Marketing, Customer
Support, Operations), and client-facing analytics for merchants — including SLA
tracking across 3, 000+ support tickets, ad performance across 5 channels,
finance reporting that cut manual work by ~6 hours/week, and productivity
reporting across 30+ team members.
• Designed and implemented CDC pipelines using PostgreSQL logical replication
(publications, subscriptions, replication slots) for incremental sync, alongside
full-refresh ETL pipelines with atomic shadow-table swap for zero-downtime
loads — integrating two source PostgreSQL databases (core and order) and 3+
external APIs (CRM, ad platforms, support tools), with raw data archived in S3
and loaded into a centralized analytical PostgreSQL warehouse.
• Modeled the data warehouse using dbt with a medallion architecture
(bronze/silver/gold) and star-schema fact/dimension tables to power analytics
and dashboards across teams.
• Orchestrated production data pipelines with Mage AI as the primary workflow
tool, managing dependencies, scheduling, and monitoring across the warehouse.
• Automated data and operational workflows using AWS Lambda, Zapier, n8n, and
Cloudflare Workers in an event-driven, serverless architecture, eliminating ~10
hours of manual reporting per week and shifting data delivery from daily batches
to near real-time.
• Built and deployed a production AI-powered customer support agent on n8n that
triages and answers inbound customer queries automatically, escalating only
complex cases to human agents — freeing the support team to focus on higher
impact issues.
• Developed a production AI-powered lead generation agent on n8n that
automatically identifies and qualifies prospects from multiple data sources,
supplying the sales team with a continuous top-of-funnel pipeline.
• Architected and shipped a customer support ticketing and voice support channel
with a bilingual (Saudi Arabic / English) AI voice agent now live across all
merchants on the platform, handling real merchant calls 24/7 (predominantly
Arabic). Designed the four-layer architecture: Retell AI for telephony and speech,
GPT-4.1 for language understanding, a vector-indexed base of 60
support articles for self-service answers, and n8n webhooks for backend lookups,
HubSpot ticket creation, and transcript logging to Google Sheets.
- Company industry:
- Internet & E-commerce
- Job role:
-
Information Technology
Personal Website