Software Engineer at Amazon·Austin, Texas

Hey, I'm Rob. I prototype and build software for data-heavy planning, markets, and measurement.

I like taking ambiguous problems and making them concrete. Usually that means talking to customers, digging through data, finding where a system is breaking down, and building a fix that holds up in practice.

$105M+
Annualized savings delivered at Amazon
99%
Runtime reduction on simulation workflows at Amazon
130M+
Kalshi trades analyzed in independent research
10K+
Peak TPS monitored on Quai at Dominant Strategies
01

About

Most of my work has been in backend and data-heavy systems: planning models, optimization tooling, network observability, and prototyping new ideas to evaluate potential impact and feasibility.

Outside of work, I spend a lot of time with prediction-market data and trading tools. I like projects and deep dives where I can start with a hypothesis, test it against real data, and be honest about whether it holds up.

Languages

  • Python
  • Java
  • SQL
  • Rust
  • C++
  • C
  • JavaScript
  • Scala
  • HTML/CSS

Tools

  • Pandas
  • NumPy
  • SciPy
  • DuckDB
  • Jupyter
  • AWS
  • Redshift
  • SQLite
  • Redis
  • Docker
  • Git
  • PyTorch
  • TensorFlow

Concepts

  • Market microstructure
  • Time-series analysis
  • Probability
  • Networking
  • Optimization
  • Quantitative analysis
02

Experience

Amazon Inc.

Austin, Texas
Software Engineer II — Supply and Resource PlanningOct 2025 – Present
Software Engineer I — Supply and Resource PlanningFeb 2024 – Oct 2025

I work on driver procurement and scheduling systems, mostly around planning models, optimization, and tools used by operations teams in North America and Europe.

  • Led development of an optimization system used by cross-functional planning teams, reducing runtime by 99% and enabling $85M in annual savings across NA/EU operations.
  • Partnered with operations and business stakeholders to launch a scalable onboarding framework, increasing planned weekly volume by 20%.
  • Collaborated with research and business teams to evaluate and deploy cost-modeling improvements projected to save $19M annually.

Dominant Strategies

Austin, Texas
Software EngineerAug 2023 – Feb 2024

Worked on Quai's core protocol team, with a focus on network observability and peer-to-peer messaging performance.

  • Built Quai-Stats, a network observability platform that automated collection and visualization of live network metrics across roughly 300 nodes — throughput, block and consensus timing, and peer health.
  • Reimplemented peer-to-peer messaging on the libp2p stack, measuring and reducing end-to-end network latency while improving message-delivery reliability across the sharded network.

Neuraflash

Atlanta, Georgia
Software Engineering Intern — AI & Einstein Solutions TeamMay 2021 – Aug 2021

Worked on Salesforce chatbot projects, including Getty Images' iStock case and sales bot, plus internal tooling for JSON package generation.

  • Increased customer satisfaction by 30% on average through chatbot maintenance.
  • Automated JSON package generation, saving the sales team roughly 2 hours per week.
03

Projects

Prediction Market Execution & Analytics Platform

Feb 2026 – Present

Personal research and trading project around prediction markets, covering data ingest, calibration research, live market monitoring, and execution tooling.

  • Ingested and queried 130M+ Kalshi trades with DuckDB to study market calibration; surfaced systematic mispricings by category, volume, and time-to-close.
  • Ran lead-lag analyses across NBA, EPL, and MLB markets, joining live game feeds (Statcast) with market prices to identify exploitable reaction latency.
  • Built a Rust execution engine consuming real-time market data over REST/WebSocket for live order entry — with position limits, risk controls, latency instrumentation, and a dashboard for discovery, order-book, and post-trade analysis.

Algorithmic Strategy Learner & Market Simulator

Mar 2025 – May 2025

Course project comparing learned and manual trading strategies using technical indicators, position limits, commission, and market-impact assumptions.

  • Implemented a market simulator with commission and market impact to compare strategies under realistic trading frictions.
  • In-sample, the bagged-tree strategy performed best; out-of-sample, the Q-Learner generalized better.
04

Education

Georgia Institute of Technology

May 2025

Master of Science in Computer Science

GPA: 4.00

Specialization: Machine Learning

Georgia Institute of Technology

May 2023

Bachelor of Science in Computer Science, Highest Honors

GPA: 3.71

Specializations: Artificial Intelligence and People (UI) · Minor: Computing & Business (Denning Technology & Management Program)