richgoodier.info
Rich Goodier

Rich Goodier

I build AI systems end to end — from a local and privacy-preserving pipeline that makes 46,000 personal photos searchable in plain language, to multi-platform text analysis and visualizations. NLP, LLMs, and computer vision, grounded in a scientist's rigor and years of leading and teaching.

Currently available

Projects

AI-Powered Personal Photo Search Engine

2025
  • Built multi-stage AI pipeline indexing 46,000+ personal photos across five metadata dimensions (captions, keywords, GPS location, facial identity, date) into a unified SQLite database queryable via natural language
  • Generated natural language captions for all photos using BLIP-2 on local GPU, then embedded them for semantic search
  • Extracted and clustered 50,000+ face embeddings for identity grouping using DeepFace and HDBSCAN
  • Built Flask web app for semi-supervised face labeling with nearest-neighbor suggestions and review workflows

RAG Chatbot over Personal Sermons and Email

2025
  • Built privacy-preserving RAG chatbot running Llama 3.2 entirely on local GPU over personal sermon transcripts and email archives
  • Paired local LLM inference with FAISS vector search for grounded question answering
  • Chunked documents with overlap and retrieved top-k relevant passages via FAISS similarity search

Sermon Corpus NLP Pipeline: Audio to Searchable Knowledge Base

2025
  • Transcribed 408 sermons (12 years of audio) using Whisper on local GPU, then extracted structured metadata per sermon via LLM API
  • Scraped and downloaded sermon audio from YouTube using BeautifulSoup and yt-dlp
  • Implemented TF-IDF analysis from scratch and via scikit-learn for per-sermon keyword extraction

Barista Name Intelligibility Metric (BNIM)

2024
  • Designed novel Barista Name Intelligibility Metric (BNIM) combining Shannon noisy channel survival, phonemic nearest-neighbor distinctiveness, length/cluster penalties, and name frequency into a weighted ensemble, ranking 1,773 names
  • Built phoneme confusion matrix from articulatory features calibrated against empirical psychoacoustic error rates
  • Computed phonemic distinctiveness via custom Levenshtein distance against 125,000+ dictionary entries

Multiplatform Narrative Analysis of Venezuela-US Relations

2025
  • Co-built multiplatform NLP pipeline analyzing 705K+ documents across Reddit, GDELT news, and TikTok spanning 13 years and 2 languages, targeting ICWSM 2026
  • Built GDELT data pipeline collecting 292K events from BigQuery and scraping 226K full-text articles with Wayback Machine fallback
  • Applied RoBERTa sentiment analysis (89% accuracy vs. VADER's 66%) and BERTopic topic modeling (15K+ topics) across all platforms
Show 13 more projects

Single-Cell Transcriptomics of Pancreatic Beta Cells in Type 2 Diabetes

2025
  • Built end-to-end scRNA-seq pipeline on HPC processing 500 GB of sequencing data through alignment, read counting, and normalization
  • Confirmed beta cell dysfunction in Type 2 Diabetes by replicating published findings on marker gene downregulation
  • Identified a mitochondrial dysfunction gene signature as a top predictor of T2D via Lasso feature selection across 16,000+ genes

Bioinformatics Concepts Interactive Simulator

2025
Live site ↗
  • Built Flask web app with five interactive bioinformatics simulations (Hardy-Weinberg, Sanger sequencing, coverage, Illumina phasing, ChIP-seq)
  • Implemented k-mer hash indexing for read alignment with scaffold assembly and coverage reporting
  • Rendered realistic Sanger gel electrophoresis visualization with log-scale fragment migration physics

BLAST Comparative Genomics Tool

2025
  • Implemented BLAST seed-and-extend algorithm from scratch, identifying 13 homologous regions >20,000 bp between human and chimpanzee chromosomes
  • Replaced O(n^2) alignment with FFT-based cross-correlation, reducing complexity to O(n log n)
  • Engineered k-mer quality filtering to suppress repeat-masked noise and improve alignment specificity

Continual Learning with Elastic Weight Consolidation

2024
  • Implemented Elastic Weight Consolidation (EWC) in PyTorch, reducing catastrophic forgetting by 3x versus SGD across 10 sequential tasks
  • Benchmarked EWC against SGD and L2 regularization baselines on permuted MNIST
  • Validated EWC's 60-70% forgetting reduction across a 27-configuration hyperparameter grid search

Machine Learning Algorithms from Scratch

2024
  • Implemented Gaussian Mixture Model with full Expectation-Maximization algorithm from scratch
  • Built t-SNE from scratch with perplexity tuning, KL divergence optimization, and early exaggeration
  • Implemented LDA topic modeling via collapsed Gibbs sampling with Dirichlet priors

Game Theory AI Suite

2023
  • Built unbeatable Tic-Tac-Toe AI using Minimax with alpha-beta pruning
  • Built fully playable 2048 with correct cascading tile movement resolution
  • Built Minesweeper AI using knowledge-based inference to derive safe cells and mines

Multiplayer Wordle Website

2023
Live site ↗
  • Built full-stack multiplayer Wordle in Flask with user authentication, server-side sessions, and a live leaderboard ranking users by win percentage
  • Solved duplicate-letter matching edge cases with a two-pass mark-and-skip algorithm
  • Designed SQLite schema for per-game statistics powering live leaderboard rankings

e-Reader Backend: Library Management with NLP Text Analysis

2023
  • Designed Book/Library class hierarchy with ISBN metadata retrieval, reading progress tracking, and CSV import/export
  • Built NLP pipeline with POS-aware lemmatization for word frequency analysis and theme extraction
  • Implemented cross-book search by title and by quote across all pages

Cancer Deaths Visualization Dashboard

2024
  • Built interactive Flask web dashboard for exploring global cancer deaths data (5,994 records, 190 countries, 29 cancer types, 1990-2019) with three-dimensional dynamic filtering by country, year, and cancer type
  • Implemented adaptive plot routing that selects scatter, bar, or 3D visualization based on user filter selections
  • Designed ETL pipeline with data cleaning and dynamic filter population

Wiki Encyclopedia (Django)

2023
  • Built Django encyclopedia app with file-based Markdown storage using Django's default_storage abstraction, enabling CRUD operations on entries without database models
  • Implemented Wikipedia-style search with exact-title redirect and substring fallback across all stored entries
  • Rendered Markdown to HTML server-side, keeping articles human-editable as .md files

Real-Time NFL Score Tracker

2023
  • Built real-time NFL score tracker in Python that scrapes theScore.com with BeautifulSoup4 and regex, tracking all weekly games with 60-120 second refresh cycles
  • Parsed live game data from raw HTML using regex pattern matching
  • Implemented score-change inference logic that classifies scoring plays by point differential (field goal, touchdown, extra point, safety/2-pt conversion) from successive polls

Hash Table Spellchecker in C

2023
  • Implemented hash table spellchecker in C with separate chaining, supporting a 143,000-word dictionary at O(1) average lookup
  • Built character-by-character parser handling contractions, digits, and word length edge cases
  • Validated against 26 reference texts totaling 5.6M words with microsecond-precision benchmarking

Google Search Front-End Clone

2023
  • Built three-page Google Search front-end clone (simple, image, advanced) in HTML5/CSS3 with Bootstrap 5, submitting queries directly to Google's live search endpoints via GET forms
  • Implemented advanced search page mapping four labeled inputs to Google's URL parameters (as_q, as_epq, as_oq, as_eq) for Boolean-style query filtering
  • Replicated Google's rounded search box UI and dual-button layout ("Google Search" / "I'm Feeling Lucky") using CSS border-radius and Bootstrap utility classes

Work Experience

Data Scientist

10/2025 – present

Zeus Learning · Remote

  • Built BERT classifiers to auto-tag K-12 question stems across four educational taxonomies, achieving high accuracies on highly complex taxonomies
  • Designed a multi-agent system to generate tens of thousands of reliable synthetic training examples, replacing insufficient provided datasets
  • Tackled classification across hundreds of fine-grained educational standards, the project's most challenging taxonomy

Freelance AI Developer

1/2023 – present

GliderFlow AI · Lexington, MA

  • AI-Powered Book Recommendation System: Developed an AI assistant for an independent bookstore (client: inklingsbookshoppe.com) that guides customers to new books based on natural language input and reading preferences. Integrated LLM-based query handling with metadata-based filtering.
  • Church Administration Automation (receiptbot): Built automated receipt processing pipeline using Vision API to extract financial data from PDFs and classify expenses against 30+ church accounting codes
  • Added semantic search to retrieve similar past receipts as few-shot examples, improving extraction accuracy with a vision-to-text fallback

Computer Science Teacher

1/2023 – 1/2025

New Covenant School · Arlington, MA

  • Taught a weekly course on digital literacy and computer programming to all 1-5 graders.
  • Created a 33-week proprietary curriculum covering computer hardware, binary code, Internet architecture, digital responsibility, and programming logic.

Senior Pastor

9/2010 – 6/2022

Mount Hermon Baptist Church · Durham, NC

  • Led a congregation of 150 and a staff of 7.
  • Developed strategies to save $30K annually, cut administrative overhead by 50%, and implemented a new church vision, including a prison ministry.
  • Organized weekly worship services and Bible studies while enhancing leadership collaboration.
Show 4 more roles

Junior Specialist

9/2005 – 6/2007

University of California at Davis · Davis, CA

  • Conducted research on the Potassium-Chloride Co-transporter 2 (KCC2) protein, including molecular characterization and physiology studies in the lab of John Payne, Ph.D.
  • Developed a novel antibody for KCC2.
  • Co-authored one research paper.

Hematology Technician

1/2005 – 11/2005

IDEXX Laboratories · West Sacramento, CA

  • Operated and maintained Bayer ADVIA 120 hematology machines for testing mammalian bloods.
  • Produced and stained slides for manual differential; prepared manual hematology tests.
  • Ensured accuracy in tests including CBC, reticulocyte count, and other hematology parameters.

Research Assistant

5/1997 – 7/1997

National Institutes of Health · Bethesda, MD

  • Studied Helicobacter pylori within the Institute of Diabetes and Digestive/Kidney Disease under Dr. William Coleman.

Research Assistant

9/1996 – 5/1997

University of Maryland Baltimore County · Baltimore, MD

  • Conducted laboratory experiments on Saccharomyces, studying RNase MRP, under Dr. Janice Zengel
  • Main tasks included setting up polymerase chain reactions and gel electrophoresis.
  • Presented research findings to faculty and students; authored a journal article.

Volunteer

Board Member / Webmaster

10/2020 – present

Christians Caring for Creation · Durham, NC

  • Contributing to the strategic direction of the non-profit organization as a board member.
  • Managed the social media presence and website ChristiansCaringForCreation.org (2020-2024).

Education

Northeastern University

9/2023 – 5/2026

Masters in Data Science GPA: 4.0

Coursework in machine learning, data mining, artificial intelligence, deep learning, bioinformatics, and visualization.

Duke Divinity School

9/2007 – 5/2010

Masters in Divinity

Coursework in leadership, administration, and financial management.

Harvard University

9/1997 – 6/2001

AB Cum Laude in Biology

Advanced coursework in statistics, mathematics, and neurobiology.

Certifications

AWS Certified AI Practitioner badge

AWS Certified AI Practitioner

Amazon Web Services