Jason Lotzer · Data Engineering Portfolio

Selected Projects

Core Metric System for Games (DAU / Retention / Engagement)

Designed a single source of truth for business and operational metrics with dimensional models (facts/slowly-changing dimensions) and consistent metric contracts. Implemented backfill-safe partitioning and governed table exposure for product analytics and ML.

Explained and versioned metrics
Backfills without downtime
Dashboards auto-updated via contracts

Real-time Telemetry → Lakehouse

Built Flink/Spark streaming topologies with event-time windows, watermarks, and late-data handling to land analytics-ready tables (Delta/Iceberg). End-to-end lineage and data quality checks shift DQ left into CI/CD.

Exactly-once sinks
Sub-minute freshness
99.95% pipeline SLA

CI/CD for Data + Observability

Git-driven orchestration with unit + schema + expectation tests, contract enforcement, and canary deploys. DataDog/Prometheus alerts, lineage, and auto rollbacks trim MTTR.

500+ assertions in CI
<15 min rollbacks
0 P1s post-launch

Live Core Metrics Demo (Synthetic)

Tiny simulated event stream. Adjust inputs and watch DAU/MAU and cohort retention respond in real time.

New users / day 200

First‑day retention 0.60

Decay half‑life (days) 10

Comeback rate 0.03

Day: 0

Model: P(active on day d) = FDR × 2^-d/HL + comeback

DAU

0

MAU (30d)

0

Day‑1 Retention

0%

Day‑7 Retention

0%

Exactly-Once, Late Events & Watermarks (Synthetic)

Out-of-order events stream in with duplicates. The processor tracks a watermark (max event-time seen minus allowed lateness) and emits exactly once per event ID per window. Late events (older than watermark) are dropped.

Rate (events/s) 120

Out-of-order jitter (s) 5

Allowed lateness (s) 10

Duplicate probability 0.08

Window size (s) 10

WM window: 0

Raw arrivals (current window)

0

Exactly-once output

0

Duplicates dropped

0

Late dropped

0

Experience

2022 → Now

Senior Adobe Commerce Cloud Developer — Primo (water.com)

Led platform-scale data integrations and product analytics in a high-traffic commerce environment: built feeds (Google Merchant), API services (Salesforce location search, 38k+ locations), and telemetry pipelines enabling merchandising and growth experimentation. Stack: Linux, PHP/Node, Python, MySQL, Elasticsearch, MongoDB, Kubernetes, Git.
2011 → 2022

Full-Stack Web Developer — ICM Corporation

Operated 60+ e‑commerce & B2B sites across mixed infra. Built data flows between Magento, Odoo, Akeneo, and internal systems; shipped search/analytics (Elasticsearch, GA/Tag Manager) and automated content pipelines. Introduced Git/Jenkins CI and containerized services.
Earlier

Engineering & Design Roles

Graphics + ASP.NET storefronts; SQL Server work; hardware builds. Foundation that informs pragmatic, product-minded engineering.

Why Netflix Games

Own Core Metric Systems: I build governed metric layers with explainable definitions and contract‑driven dashboards.
Data Modeling Architecture: Dimensional designs with SCDs, aggregations, partitioning, and backfill strategies.
High‑Reliability Pipelines: Spark/Flink for batch + streaming; event‑time partitioning, late data, CDC, upserts.
Governance & Quality: Lineage, metadata, and expectation suites (accuracy, completeness, validity, consistency).
Strategic Collaboration: Translate DS/PM goals into scalable data products; ship dashboards and feature tables.
Technical Leadership: Reduce toil, pay down debt, automate checks, mentor peers, and champion best practices.

Let’s Talk

Email jasonlotzer@gmail.com · Hartford, WI · LinkedIn