Michigan-based · Open to US-remote
Founder of Hello Hoku, where I build practical tools for special education and care delivery. I also work on geospatial ML for conservation with transparent, evaluation-first workflows.
MSc in Computer Science — dissertation on predicting orangutan poaching risk (Borneo) using Google Earth Engine + geospatial ML.
MSc in Developmental Psychology - The relationship between linguistic skills and linguistic prediction
BSc in Psychology
Founded Hello Hoku — direct intervention for children with disabilities; shipping small tools that save clinician time.
-
Poaching Risk Maps (Borneo) — Risk surfaces from forest-loss + access proxies + conservation context.
scikit-learn · geopandas · rasterio · GEE
➜ https://github.com/xxvfotia/poaching-risk-maps -
IEP Assistant — Deterministic JSON → IEP draft (Markdown) with SMART goals + red-flag checklist.
python · templates · education
➜ https://github.com/xxvfotia/iep-assistant -
Session Note Summarizer — Weekly digests + tag extraction with precision/recall/F1.
nlp · evaluation · python
➜ https://github.com/xxvfotia/session-note-summarizer -
Trainer Scheduler Helper — Greedy trainer-to-session assignment (availability + capacity) → schedule CSV.
scheduling · ops · python
➜ https://github.com/xxvfotia/trainer-scheduler-helper
Core ML & Data: Python, NumPy, pandas, scikit-learn, XGBoost/LightGBM, model evaluation (ROC/PR-AUC, calibration), error analysis, cross-validation, spatial CV (block/leave-region-out)
NLP & LLMs (lightweight): text preprocessing, keyword/rule baselines, prompt design, evaluation (precision/recall/F1)
Geospatial: GeoPandas, Rasterio, Shapely, GDAL/OGR, QGIS, Google Earth Engine, CRS/projections, tiled map outputs
MLOps & Reproducibility: Git/GitHub, conda/pip, .gitignore, deterministic baselines, experiment logs, README-driven docs, data/version hygiene
APIs & Apps: FastAPI (simple inference APIs), Streamlit (light UIs), Markdown/CSV automation
Data Engineering (practical): SQL (PostgreSQL), CSV/Parquet, batch jobs, basic orchestration (cron/Makefile)
Packaging & CI: Docker (containerizing small services), GitHub Actions (lint/test), pytest (unit tests)
Cloud (practical): AWS S3/EC2 (files + jobs), GCP BigQuery (tabular), Google Earth Engine (raster/time series)
Ops & Analytics: scheduling heuristics, simple optimization (greedy/first-fit), reporting dashboards (matplotlib)
Privacy & Compliance (education): de-identification/PHI redaction, HIPAA/FERPA awareness, safe sharing of anonymized artifacts
Languages: English (fluent), Chinese (fluent), Malay (fluent), Korean (fluent), French (learning)
US-remote roles where I can ship applied ML and clinician-facing tools with measurable impact—and keep results transparent (clear baselines, evaluation, limitations).
Email: tahaeffy@gmail.com · LinkedIn: www.linkedin.com/in/effytaha · Location: Michigan, USA