George Liu

About

I’m a software engineer based in Northern Virginia. I care about precise systems, clear data models, and how information actually moves — through pipelines, markets, and networks.

Contact

Background

I spent several years on the Knowledge Graph and Data Platform team at FINRA (2022–early 2026), working with Neo4j, Apache Spark, Dagster, and AWS to build entity-resolution and lineage-aware pipelines. That work shaped how I think about correctness, scale, and operational discipline.

I’m currently focused on independent work: systems programming and low-latency software on Linux, quantitative finance tooling, prediction markets, and practical automation — including tools built with modern agentic CLI workflows.

I hold a B.S. in Statistics with a Computer Science minor from the University of Wisconsin–Madison. Outside of engineering I play competitive tennis (USTA league and social doubles) and take fitness and personal development seriously.

Timeline

  1. 2026–present

    Independent engineering

    Systems programming, quantitative tooling, prediction-market experiments, and automation.

  2. 2022–early 2026

    FINRA — Knowledge Graph / Data Platform

    Entity resolution, data lineage, Neo4j, Spark, Dagster, and AWS-backed production pipelines.

  3. Earlier

    University of Wisconsin–Madison

    B.S. Statistics · Computer Science minor.

Expertise & interests

Areas of work

  • Data platforms and batch/stream pipelines
  • Knowledge graphs, entity resolution, and lineage
  • Spark (Scala / PySpark), Python, Neo4j / Cypher
  • AWS (S3, EMR, EC2, ECS, RDS), Docker, CI/CD
  • Growing focus: C++, Linux, performance engineering

Current interests

  • Prediction markets and information flow
  • Statistics, trends, and social networks
  • Systems performance and low-latency design
  • Competitive tennis and fitness

Working style

Precision over theater

Prefer correct models, measurable behavior, and honest uncertainty over impressive demos.

Systems thinking

Look at interfaces, failure modes, and data flow — not only the happy path.

Disciplined iteration

Ship small, learn quickly, keep the design simple enough to explain.

Curiosity with rigor

Follow interesting questions, then pressure-test them with evidence.