Senior Data & Platform Engineer. I design, build, and operate production data platforms end-to-end — ingestion, orchestration, storage tiering, and the data quality and governance layers that make them trustworthy.
PhD in Intelligent Transportation Systems (University of Toronto). Currently leading data platform delivery on a $40B+ infrastructure program. Toronto, Canada.
Production data systems where the interesting problems are architectural, not cosmetic: medallion-layer pipelines, hot/cold storage tiering, OLTP/OLAP separation, data quality and lineage frameworks, and migrations of legacy estates to modern architectures without disrupting the people downstream who depend on them. I care about access-pattern-driven design, observability, and knowing when not to add complexity.
QuantumLane — a production
real-time transit data platform, live and operated on a
single small box for under CAD $20/month.
Ingests TTC GTFS-Realtime feeds, splits storage across a hot tier (PostgreSQL/PostGIS)
and a partitioned cold tier (S3/Parquet, Hive-style keys), and serves live and
historical queries through deliberately separate access paths behind a public FastAPI
layer. Memory-bound streaming (server-side cursors, COPY from open archives) loads
4M+ row datasets within tight memory; every architectural decision is captured in an ADR.
Python · Dagster · PostgreSQL/PostGIS · S3 · FastAPI · Docker
SUMO FPGA Case Study — a
complete FPGA acceleration pipeline for a traffic simulator on AWS F2, and the honest
postmortem of why it doesn't beat the CPU. A working HLS kernel, a custom PCIM DMA
engine, and bit-exact CPU/FPGA results — capped at ~1.5x by Amdahl's Law and PCIe
transfer cost, with the analysis worked out in full. A study in measuring rigorously
and reporting the negative result honestly. (write-up)
HLS · SystemVerilog · AWS F2 · Performance analysis
- Prolog Data Catalog — an experiment in using logic programming for data lineage, impact analysis, and governance queries over a medallion-architecture metadata model.
- SeeTheMath — a React Native / Expo universal app (native Android + static web from one codebase) that teaches middle-school math through interactive visuals. Built for my own kid; shipped to the Play Store.
My data-engineering work at the City of Toronto is public, merged into the city's
open-data platform: custom Airflow operators and CI/CD in
bdit_dag_utils and
bdit_data-sources. The
vehicle-for-hire pipeline I built now feeds a public Toronto Open Data dataset.
Pipelines & orchestration — Apache Airflow, Dagster, Python, SQL, dbt, Spark/PySpark Storage & modeling — PostgreSQL/PostGIS, Snowflake, dimensional modeling, medallion architecture, hot/cold tiering Quality & governance — data quality frameworks, lineage, metadata, observability Cloud & ops — AWS, Docker, CI/CD (GitHub Actions), Linux
Open to Senior / Staff Data Engineer conversations.