Daniel Bove

Principal cloud architect and software engineer focused on AWS platform systems, distributed data architectures, and developer infrastructure. I build scalable systems, reduce operational complexity, and turn platform investments into measurable engineering and business impact.

Resume

Cloud and Software Architect

I design systems and write the code that runs them. View or download my current resume (PDF).

GitHub Project

Doubleday

Doubleday is a serverless data platform for MLB pitch analysis, designed and built end-to-end on AWS. It ingests pitch-by-pitch Statcast data from Baseball Savant, processes it through a medallion-architecture lakehouse (S3, Apache Iceberg, Athena), and serves precomputed analytical views through a low-latency REST API backed by DynamoDB.

A cross-platform app built with Expo React Native lets users explore pitcher repertoires, visualize pitch movement, and discover similar pitchers across seasons. The full system, including ETL pipeline (Step Functions, Lambda), API layer (API Gateway, Cognito, Stripe), infrastructure (Terraform, GitHub Actions CI/CD with OIDC), and frontend, is designed around idempotent partition-safe processing, bronze-layer caching for cheap reprocessing, and a serverless-first architecture with no long-lived compute.

Read more...
View on GitHub

GitHub Project

RotoHero

RotoHero is a real-time fantasy baseball draft assistant powered by a multi-agent LLM architecture. Four specialist AI analysts covering value, positional need, draft urgency, and positional scarcity run as independent background threads, reactively re-evaluating recommendations as picks happen across the league.

The system uses a blackboard-style shared discussion model where agents accumulate and refine strategy asynchronously, with debounced triggers and rolling context windows to manage API costs. A strategist agent synthesizes the live discussion on demand into a single pick recommendation. The pipeline supports three execution modes (parallel fan-out, multi-round debate, and fully reactive) behind a unified protocol interface, with strategy algorithms including Monte Carlo draft simulation and statistical talent-cliff detection. Built with LangGraph, LangChain, and Python, with full type checking, 172+ unit tests, and a Rich-powered interactive REPL.