Java · Spring Boot · Distributed Systems · Data Platforms · Azure
Senior backend engineer with 10+ years building Java, Spring Boot, and Kubernetes microservices for telecom and finance.
Data Engineer · Toronto
Hired to modernize data-intensive paths on AKS. Owned the upload pipeline and source-onboarding experience end to end — from batch windows that stalled operators to config-driven loaders and latency outcomes a hiring manager can probe in an interview.
OWNED
End-to-end ownership of the file-upload path and API source onboarding for platform consumers.
BEFORE → AFTER
Multi-minute batch upload windows and week-scale API source onboarding → 60% ↓ upload latency and 40% ↓ feed onboarding via a config-driven API Source Loader, with OTel → Datadog so regressions surface in ops.
INTERVIEW
"How did you keep the loader config-driven without turning every new feed into a deploy?"
Java · Spring Boot · AKS · Helm · OpenTelemetry · Datadog · Spark/Databricks · Iceberg/Delta
Software Developer · Toronto
Built a Spring Boot recovery console operators actually use: detect inconsistencies, emit actionable reports, and remediate without opening an engineering ticket for every incident — cutting operational overhead instead of adding queue noise.
OWNED
Designed and delivered ART for ops inconsistency detection and remediation workflows.
BEFORE → AFTER
Manual chase of inconsistent billing/ops data → 25% ↓ operational overhead with a detect → report → remediate loop.
INTERVIEW
"What's the difference between a recovery tool operators trust and one they ignore?"
Java · Spring Boot · Thymeleaf · AWS · Pub/Sub
Systems practice (self-hosted)
Long-running self-hosted platform used to rehearse the same habits as production backends: named blast-radius boundaries, ingress + identity at the edge, and observability you actually look at. The Selected Builds card shows the fleet topology; this study is about the operating posture.
OWNED
Sole operator — design, secure, and run the fleet end to end.
BEFORE → AFTER
Single-box ad-hoc containers → named hosts, plane separation (edge / apps / data), and fleet monitoring you can query when something breaks.
INTERVIEW
"What did running your own edge teach you that a cloud tutorial did not?"
Traefik · CrowdSec · Authelia · Pangolin · Docker · Beszel · Dozzle · Homarr
Self-hosted platform
Six-host fleet with Traefik/CrowdSec/Authelia at the edge, app nodes, media as a workload line, and Beszel/Dozzle/Homarr across the board — not "I run Plex."
Sole operator of a six-host edge / apps / data fleet with shared monitoring.
BEFORE → AFTER
Ad-hoc containers on one box → named hosts, ingress + authz, observability, and clear blast-radius boundaries.
TALKING POINT
"Walk me through how you'd isolate a compromised edge node without taking the media or app planes down."
TD Securities · via Endava
Owned the path from slow batch uploads and week-long source onboarding to config-driven loaders and measurable latency wins — Spring on AKS, with observability wired for real ops.
End-to-end ownership of upload path + source onboarding UX for platform consumers.
BEFORE → AFTER
Uploads stuck in long batch windows; new API sources took weeks to land → 60% ↓ upload latency, 40% ↓ feed onboarding time via config-driven API Source Loader.
TALKING POINT
"How did you keep the loader config-driven without turning every new feed into a deploy?"
Amdocs Canada
Built a Spring Boot recovery console that detects inconsistencies, produces actionable reports, and supports remediation — cutting operational overhead instead of adding another ticket queue.
Designed and delivered ART used by ops for inconsistency detection and remediation workflows.
BEFORE → AFTER
Manual chase of inconsistent billing/ops data → 25% ↓ operational overhead with detect → report → remediate loop.
TALKING POINT
"What's the difference between a recovery tool operators trust and one they ignore?"
Full-stack product surfaces — secondary to platform work above.
Multi-tenant admin dashboard for managing products, orders, and analytics. Built with Next.js, integrated with Stripe for payments and Clerk for authentication.
Modern e-commerce storefront with product catalog, shopping cart, and checkout flow. Headless CMS architecture consuming admin API.
It's 3 AM. A customer paid, the orders row is in your database, and the OrderCreated event never reached Kafka. The warehouse never shipped. The email never sent. Somewhere in your service there is code that looks like this: @Transactional public Order placeOrder(PlaceOrder …
YNAB is $14.99 a month — $180 a year — for envelope budgeting. That's a subscription to be told where your own money went. Actual Budget is the open-source clone: same "give every dollar a job" philosophy, local-first so it works offline, and a tiny server that …
The client times out waiting for your POST /orders. It retries. Your server processes it twice. The customer gets charged twice, and your support queue gets a new entry titled in all caps. Stripe solved this a decade ago with one header: Idempotency-Key. The client generates a …
My monitoring stack used to be five containers: Prometheus, Grafana, node-exporter, cadvisor, and an uptime checker — plus a paid UptimeRobot plan for the "is the house on fire" pings. Total maintenance burden: a weekend to set up, and a Sunday every few months when …