A cross-section of my work history: pipelines, migrations, extraction, and analysis
Filter by category
33 projects I built a reusable asyncio-based API toolkit for data migration from one legal CRM to another, and have used it to execute 9 migrations to date. I extracted and cleaned PDF text into labeled training chunks, and enriched the data with question-answer pairs generated via the OpenAI API. I executed semantic text matching using embeddings and scoring to surface distinctive passages. I built an automation to create SEO-focused product image alt text for thousands of images. I built a semantic ranking system using ML with configurable similarity, scoring, and exclusion models. I extracted 4,000 records from old PDF book scans with uneven OCR'd text in a 3-phase process, then delivered them with a CLI lookup tool. I conducted an analysis of longitudinal intervention data that delivered a client-ready report with figures in a Jupyter notebook. I cleaned, standardized, and deduplicated 320,000 contact records, then segmented them into deliverable CSV files. I combined multi-year research datasets into a canonical dataset, then generated random samples of control and experimental groups. I cleaned and matched 10.8 million records using PostgreSQL and refined logic to improve accuracy. I built an API-based migration toolkit and used it to migrate data from one CRM to another. I built a tool to parse purchasing department chat logs, extract offers made by outside salespeople, and summarize them in a financial spreadsheet. I built a Google Sheets automation that downloads daily shuttle trip spreadsheets, cleans & colorizes them, and uploads them as new fixed sheets. I mined data for ~4,000 entities in two game systems to create a model based on percentile mapping that converts entities from the old system to the new one. I built a Django backend that implemented a RESTful API for a discography website. I processed 107 MB of XML public records to extract corporate officer data and deliver it as a lead list spreadsheet. I built an automation, intended to be run daily, that handles pixel-data collection, enrichment, and delivery to each customer's preferred destination. I built a Gmail API tool that extracts inquiry data into cleaned CSV or Excel files. I optimized chess analysis software. I improved performance, data retrieval, and parsing accuracy. I optimized a Chinese text processing algorithm across millions of database rows, reducing runtime from 19 hours to 2.5 hours. I built a sync tool to map data between a legacy student info system and its modern replacement to allow them to operate in parallel. I built a tool for transforming automotive fitment data into structured catalogs, which also identifies missing and inconsistent source records. I analyzed law firm call logs to assess staffing needs and evaluate agent performance. I built an OCR tool that automatically tunes parameters using grid search to maximize accuracy for each document set. I conducted an analysis of the results from an at-home sleep study, producing a report with supporting visualizations. I built API tools to retrieve chassis registry, safety, and compliance data, parse it, and deliver it as JSON. I built an AWS-hosted incident intake and triage tool that automatically scores and routes records. I built a pipeline for reconciling multi-source marketing data and delivering a unified dataset. I built an AWS-hosted marketing pipeline with staged processing, reproducible execution, and detailed reporting. I built a FastAPI-based lead intake service that runs on webhook submissions; it scores, classifies, and processes each record and dispatches it via Zapier. I built an ETL platform that ingests files via webhooks, and supports multiple processing pipelines. It features replay, checkpointing & resume. I built an inventory-processing pipeline based on the Shopify Admin API. It generates risk forecasts, reorder recommendations, and reports in Google Sheets. I built an ETL pipeline for validating data through automated scanning, issue tracking, and workflow integration.Reusable Migration Toolkit for a Legal CRM
Data Extraction from Uneven Book Scans to Create a 1.1M-Token Training Corpus
Hybrid Embedding and BM25 Semantic Matching Pipeline
Multimodal LLM Alt-Text Generation for an Ecommerce Catalog
Semantic Search & Ranking Engine Built on Transformer Embeddings
Fitting a Distribution-Mapping Model to Convert a 4,000-Record Corpus
Analysis and Reporting for Longitudinal Intervention Study
Bulk Contact Data Cleaning and Segmentation
Combining Segmented Dataset Across Several Years into One Spreadsheet
Data Matching & Deduplication Across 10.8M Rows
Data Migration via API Integration
Data Mining for Offer Data from Procurement Chat Logs
Google Sheets Reformatter Program
Mined Legacy Game System Data to Build Model for Converter
RESTful API Using Djagno
XML Lead Generation with LLM-Normalized Title Data
AWS Automation for Enriching Pixel Data
Gmail Messages Data Miner
Multithreaded Chess Analysis Optimization and Other Fixes
Refined Python-MySQL Algorithm Performance Across 2.8M Rows
Sync Student Data Between Legacy System and Its Replacement
Auto Parts Data Formatter
Data Analysis of Call Logs
Self-Tuning OCR with ML Parameter Search
Sleep Study Results Analysis and Report
API Query Tools for Chassis Tracking Compliance
AWS-Based Incident Management + Response System
Campaign Reconciliation Pipeline with Multi-Source Matching
Dockerized Marketing Analytics Pipeline with Staged Execution and Run Inspection
Lead Intake API with Zapier Automation
Multi-Pipeline Ingestion Platform
Shopify Inventory Reporting Pipeline with Google Sheets
Airtable- and Zapier-Based Data Integrity Monitoring
Looking for something specific?
What I do →