Interview portfolio
From discovery
to working software.
Customer engagement, hands-on engineering, and the motivation behind how I work.
“Describe a recent AI or data application you personally built. What did you code, which APIs and data platforms did you use, and how did you demonstrate the solution?”
Personal project / End-to-end ownership
HeiRenny
A sports data foundation, a conversational research application, and a focused X research service.
Problem statement
Turn fragmented sports data into answers people can inspect and trust.
Researching players and matchups requires combining statistics, schedules, injuries, news and other context from multiple providers. Different player identifiers and data conventions make those sources difficult to join reliably. An AI research interface also needs a dependable data foundation and explicit controls on how it queries and interprets that data.
Project context
I built HeiRenny end to end to connect that data foundation to a usable research experience. My work spans Python ingestion with dlt, SQL and dbt modeling in Snowflake, curated reporting in Snowflake Postgres, and a FastAPI backend with a React/TypeScript frontend. A LangGraph agent uses typed tools to research questions and return supporting tables and visuals. The same reporting foundation supports the X service I am building for deployment, where Renny will answer real users’ sports questions. I am currently developing and testing that workflow in the local Question Lab.
01 / Data engineering architecture
Data engineering pipeline
Sports information arrives from several providers in different formats. The pipeline organizes it, works out which records refer to the same player, and checks the results before making them available to the application and X service. Uncertain player matches are kept for review.
Browse all data engineering components
Read the interview answer +
A recent example is HeiRenny, a sports data and AI application I built end to end. It brings together player statistics, schedules, injuries, news and other context so users can research players and matchups through a conversational interface and inspect supporting tables and visuals.
My work spans Python ingestion pipelines using dlt, SQL transformations using dbt in Snowflake, curated reporting data published to PostgreSQL, a Python FastAPI backend, and a React/TypeScript frontend. I integrated sports-data and research APIs and built an agentic workflow with LangGraph and LLM APIs to connect natural-language questions to typed tools and analytical queries.
The most interesting engineering work was making the data and AI behavior trustworthy. I implemented AI-assisted player identity resolution with deterministic checks and an audit trail for uncertain matches. I also constrained the research agent’s database access with SQL parsing, approved-table restrictions, read-only queries and result-size limits.
I have not yet given a formal demonstration of HeiRenny. I would be happy to walk through the application, code and data pipelines if requested. It is my clearest example of these capabilities in one project; I can also share relevant examples from my professional work across other use cases.
Implemented architecture, not a live execution trace. The X service is in development and testing before deployment, with the goal of having Renny answer questions for real users. Deployment activation for the separated repositories is being handled through migration.