Home

Projects

Selected projects

EIA Electricity Data Pipeline

Team training capstone at Revature · major pipeline contributor · Mar 2026 – Apr 2026

Ingests U.S. EIA electricity data into Snowflake RAW, SILVER and GOLD layers, orchestrated by Airflow and validated with dbt.

  • Python
  • SQL
  • Apache Airflow
  • Snowflake
  • dbt
  • Streamlit
  • Docker Compose
  • pytest
  • Major contributor across pipeline architecture, ingestion, transformation, backfills, testing and Streamlit data access.
  • Refactored Airflow into separate hourly and monthly ingest and transform pipelines with incremental ingestion and serialized backfills.
  • Added dbt source-freshness and schema tests and a pytest coverage gate.
  • Moved with the team from an earlier Spark, Kafka and MinIO design to the final Snowflake-first pipeline.

Instacart Bronze/Silver/Gold Data Platform

Team training project at Revature · broad implementation ownership · Feb 2026 – Mar 2026

Dockerized PostgreSQL Bronze/Silver/Gold platform for Instacart order data, with validation gates and Streamlit analytics.

  • Python
  • PostgreSQL
  • pandas
  • SQLAlchemy
  • Docker Compose
  • Streamlit
  • pytest
  • Had broad ownership of the platform and pipeline implementation.
  • Built a staged profile, bronze, silver, gold and validate workflow with reports, logs and data-quality checks.
  • Added Streamlit analytics for KPIs, time behavior, product performance, reorders, customers and baskets.

Other projects

US Accidents + Weather ETL (abandoned direction)

Early Revature team direction, replaced by the Instacart project · Feb 2026 – Feb 2026

Archived

Early Revature team direction combining U.S. accident and NOAA weather data before the team pivoted to Instacart.

  • Python
  • PostgreSQL
  • PostGIS
  • Docker
  • Streamlit
  • Built a resumable, checkpointed ETL split across accident, weather and enrichment steps into PostgreSQL/PostGIS.
  • Added Streamlit exploration pages to test whether the combined data produced useful analytical signals.
  • Archived after the team found the direction had weak analytical value and deliberately pivoted to Instacart.

EIA API Explorer

Solo domain-discovery prototype before the EIA pipeline project · Mar 2026 – Mar 2026

Completed

Solo Streamlit prototype used to explore the U.S. EIA electricity API and datasets before the larger team pipeline.

  • Python
  • Streamlit
  • Docker
  • REST APIs
  • unittest
  • Built a data loader and analysis pages around the EIA API, with unit tests and an analytics validation script.
  • Used the prototype to understand the domain and possible business use cases before the team designed the EIA pipeline.

Cache Replacement Algorithms Simulator

Older personal or school project · exact context and date not recorded

Early project

Older Python project comparing cache-replacement strategies including LRU, LFU, ARC and machine-learning-based approaches.

  • Python
  • PostgreSQL
  • Redis
  • Implemented a simulator covering LRU, LFU, ARC and machine-learning-based replacement strategies.
  • The original workloads and hit-rate results are not currently recorded, so the archive does not claim a best-performing strategy.

Atari Transformer RL Agent

Older personal or school project · exact context and date not recorded

Early project

Older Python/PyTorch project: a PPO reinforcement-learning agent with a Transformer encoder for temporal modeling on Atari games.

  • Python
  • PyTorch
  • Reinforcement learning
  • Transformers
  • Used PPO for reinforcement learning and a Transformer encoder to model temporal context.
  • The archive stays intentionally narrow: the original game list, scores and baseline results are not currently recorded.

Fintech Trading Platform

Older personal or school project · exact context and date not recorded

Early project

Older Python trading-platform project built with Kafka, PostgreSQL and Redis.

  • Python
  • Kafka
  • PostgreSQL
  • Redis
  • Built an automated trading-platform project using Python, Kafka, PostgreSQL and Redis.
  • Earlier wording about low latency and real-time analytics is intentionally omitted because no supporting measurements are currently recorded.