Ben Lee

Ben Lee

Software Engineer

Software Engineer with 4 years of production experience building backend APIs, ETL pipelines, and LLM-integrated systems. Shipped a production LLM agent pipeline on AWS Bedrock saving 200+ engineering hours per quarter; designed real-time data pipelines across energy and logistics; built full-stack applications end-to-end from schema to deployment. Independently shipping full-stack products — from concept to App Store — with a focus on clean product thinking and engineering craft.

Software Developer III, Paycom

Irving, TX | March 2025 – February 2026

Designed and owned REST APIs in C# (.NET Core) and PHP for high-volume payroll event processing across a multi-tenant architecture serving 20,000+ enterprise clients at 99.9% uptime.

  • Built Splunk observability layer with structured logging and alerting, reducing mean-time-to-detect across distributed client environments.
  • Shipped internal data reporting tool that cut manual ops effort by 30%.
  • Used AI-assisted development (Claude) throughout to accelerate delivery and improve code quality.

Software Engineer, Valent Partners

Dallas, TX | June 2022 – March 2025

Delivered full-stack engineering across four client engagements in energy, logistics, insurance, and enterprise AI — each greenfield, each in production.

  • Designed and deployed a production LLM agent pipeline on AWS Bedrock integrating enterprise data sources, including prompt engineering framework, agent orchestration, and output validation — saving 200+ engineering hours per quarter.
  • Firefly Energy — ETL pipeline extracting and transforming time-series energy data into a Django app with embedded Tableau dashboards for automated client-facing analytics.
  • Raad Systems — Real-time event-driven data pipeline on Azure Functions and MSSQL in C#, ingesting supply chain streams, applying transformation and scoring logic, and triggering automated risk alerting.
  • Goosehead Insurance — Node.js backend and React frontend for a DTC insurance platform; PostgreSQL schema design, carrier API integration, 22% reduction in customer acquisition cost.
  • Implemented fault-tolerant integrations with 10+ third-party APIs using retry logic and circuit breakers; reduced pipeline failure rates by 40%.

Intern, Consolidated Data Services, Fidelity Investments

Westlake, TX | June 2021 - August 2021

Supported backend automation initiatives that improved consistency and operational efficiency in data processing workflows.

  • Automated backend data workflows using Python.
  • Improved reliability through repeatable processing patterns.

Technical Expertise

Backend · AI Systems · Data · Cloud · DevOps

Primary stack is Python and TypeScript, with production experience across AI/LLM systems, ETL pipelines, and full-stack applications.

  • Languages & Frameworks: Python (FastAPI, Django), TypeScript/Node.js, React, Vite, C# (.NET Core), SQL
  • AI & Agentic Systems: LLM integration (AWS Bedrock, Anthropic), prompt engineering, output validation, agent orchestration, agentic workflow design
  • APIs & Architecture: RESTful API design, microservices, event-driven pipelines, fault-tolerant systems, retry logic, circuit breakers
  • Data & Pipelines: ETL pipeline design, Azure Synapse, PostgreSQL, MSSQL, MySQL — data modeling, transformation, query optimization
  • Cloud & Infra: AWS (Lambda, API Gateway, Bedrock), Azure (Functions, Synapse), Supabase, Docker, CI/CD
  • ML: Weak supervision (Snorkel), scikit-learn, sentence-transformers, FastAPI model serving

Selected Projects

Indie Development · 2024–Present

Full-stack independent projects shipped from zero to production, each solving a real problem with a clean product layer over non-trivial engineering.

Great Teacher Onigiri Pre-launch · great-teacher-onigiri.vercel.app

Gamified GTO poker learning app — "Duolingo for poker theory." 5-world curriculum, 91+ interactive slides, a real poker engine for Boss Battles, and a Player Profile tracking 7 live metrics (VPIP, PFR, aggression factor, and more).

  • React 19 + Vite, Supabase auth/database, RevenueCat for iOS IAP (Onigiri Pro $9.99/mo), and Capacitor for native iOS/Android packaging.
  • Freemium hearts system gates learning velocity — not content access — to reduce churn friction while monetizing power users.
  • Fixed three non-trivial production bugs: SPA 404 via Vercel rewrites, AppDelegate build failure under Xcode 16 via protocol cast, and a blank-screen race condition resolved through state machine redesign.
Live Demo GitHub Case Study →
h1nt Live · h1nt.vercel.app

AI-powered technical interview prep with a personalized study plan, a streaming "Tiger Parent" Claude Sonnet tutor, Monaco code editor for LeetCode questions, and a graded mock interview locked behind a 70+ Readiness Score.

  • Next.js 16 App Router (TypeScript), Vercel AI SDK v6 streaming, Zustand, shadcn/ui, and Claude Sonnet 4.6 as the AI backbone.
  • Smart question ordering: weak areas first → difficulty → failed review. "I Don't Know" teach mode explains before grading.
  • Fixed empty-stream bug by switching /api/tutor from toUIMessageStreamResponse() to toTextStreamResponse().
Live Demo GitHub Case Study →

Education

Southern Methodist University | May 2022

B.S. in Economics, Magna Cum Laude. Strong analytical training that complements engineering problem-solving and product thinking.

  • GPA: 3.89 / 4.0