Luke Monington — Portfolio
Data Scientist & ML Engineer — Practical AI that ships
Forecasting • LLMs • Vision • Time‑series • Graphs. 8 years across healthcare, finance, and retail.
Forecasting
Time-series, GNNs, and events-aware demand
LLM
Private RAG, fine-tuning, and edge deployment
Vision
Detection/classification for ops and QA
Automation
ETL, data quality checks, and MLOps
Featured on YouTube
I publish hands‑on walkthroughs and research explainers. These are four of the most relevant for hiring managers.
Selected Projects
AI Forecasting System for Interconnected Business Data
Deep learning with Graph Neural Networks (GNNs) to model cross-entity relationships (products, people, regions, devices) and forecast context‑aware outcomes.
ForecastingGraphsGNNE2E
Real systems are interdependent. Modeling the graph directly yields better demand, risk, and capacity forecasts.
- Nutrition/CPG → ingredients ↔ suppliers ↔ plants → demand & supply risk
- Car rentals → branches ↔ hubs ↔ customer flow → vehicle availability
- Retail → products ↔ co‑purchases ↔ categories → SKU‑level demand
Estimating Outcomes from Sequences + Static Data
Neural architecture that merges time‑ordered events with item attributes to predict numeric outcomes as a live service.
PredictionSequencesTabularMLOps
Combines what happened (sequence) with what it is (attributes) for high‑accuracy predictions.
- Logistics → time‑to‑depletion from movement history + item specs
- E‑commerce → AOV/abandonment from user sessions + product/customer features
LLM Customization & Edge Deployment
Fine‑tune compact LLMs and deploy offline on consumer GPUs/embedded devices with full training→eval→inference pipeline.
LLMEdgeFine‑tuningLatency
Low‑latency, privacy‑preserving inference without third‑party APIs.
- Support intent routing
- In‑store voice assistant
- Developer command tools
Human Activity Recognition with Pose Estimation
MoveNet + temporal CNN/LSTM to classify actions in real time on commodity hardware.
VisionPoseEdgeSafety
Gesture, safety, and task‑stage recognition without wearables.
- Warehouse ergonomics
- Queue behavior
- Gesture UI
Private Document Chat (Local RAG)
Fully local retrieval‑augmented generation with private embeddings and vector search—no data leaves the environment.
RAGPrivacyLLM
Regulatory‑friendly Q&A over sensitive docs.
- Legal case files
- Clinical protocols
- Finance audit trails
Bulk Document → Structured Data
OCR + rules + NLP to turn PDFs/Word/scans into clean spreadsheets at scale.
AutomationOCRETL
Searchable datasets and QA over previously opaque documents.
- Retail reports
- Grocery invoices
- Field notes
Predictive Modeling on Structured Data
End‑to‑end pipeline from cleaning and features to interpretable models and production hooks.
TabularXGBoostShapProd
Actionable churn/returns/LTV predictions wired to dashboards and ops.
- Marketing churn
- Returns risk
- Purchase likelihood
Time‑Series Forecasting
Neural & classical models for seasonality, events, and anomalies across any logged metric.
ForecastingAnomalyARIMADeep
Anticipate demand, detect shifts, improve planning.
- DAUs
- Energy use
- Store traffic
Computer Vision for Operations
Train detectors/classifiers for shelves, packaging, safety, and custom inspections.
VisionDetectionRealtime
Automate visual checks and reduce manual review.
- Stock levels
- Planogram QA
- PPE compliance
Automated AI QA for Manual Workflows
Find duplicates, conflicts, and missing fields across human‑entered data before it ships.
QAAutomationData Quality
Protect quality and margin while reducing tedious checks.
- Catalogs
- Briefs
- Specs
About & CV
Methodical, analytical, and hands‑on data scientist with eight years in AI. I ship forecasting, returns reduction, LTV/churn prediction, and private LLM tooling. Prior projects include distributed training of clinical LLMs on AWS, reproduction of peer‑reviewed models, and production vision systems. I like clean data paths, clear metrics, and fast feedback loops.
TensorFlowPyTorchXGBoostscikit‑learnPythonSQLAWSGCPAzureDockerKubernetesNLPComputer VisionTime‑SeriesGraphs (GNN)
Experience
Machine Learning Engineer — Artisight
Remote · Jun 2022 – Present
- Delivered ML/DL and vision services for hospitals; improved predictive automation reliability.
- Spun up 8‑GPU AWS cluster for distributed LLM pretraining on hundreds of millions of clinical notes.
- Reproduced peer‑reviewed outcomes models (mortality/LOS/readmission) with AUCs ~0.945/0.881/0.889.
Data Scientist — KPMG
New York, NY · May 2021 – Jun 2022
- Built NLP pipelines (regex, Transformers, DNNs, XGBoost) for contract extraction at >0.95 precision/recall.
- Predicted client turnover; collaborated across 20+ stakeholders to productionize models.
Machine Learning Engineer — Resource Capital Funds
Denver, CO · May 2020 – May 2021
- Automated ~10% of investment decisions via 6 ML models + GUI on AWS EC2; metals exploration targeting.
- Led small team of consultants; deployed live analytics tooling for decision support.
Business Analyst Intern — Spectrum
Centennial, CO · May 2019 – Aug 2019
- Performed statistical analysis (PCA, MCA, Odds Ratios, Chi‑Squared, regularized logistic regression).
- Presented findings to leadership; ~5% cost savings.
Data Scientist Intern — Arrow Electronics
Centennial, CO · May 2018 – Aug 2018
- Built SQL pipelines (Oracle/MySQL) and dashboards (Tableau/QlikView) to support marketing campaigns (≈8% sales lift).
- Shipped internal web apps for market intelligence and e‑commerce.
Education
B.S. in Computer Science (Minor in Math)
University of Northern Colorado — Monfort College of Business
Magna Cum Laude, GPA 3.85/4, Beta Gamma Sigma (top 8%) — Founder: AI Club & IoT Club
May 2020
Certifications
- Azure AI Engineer (AI‑102) — Feb 2022
- Azure Fundamentals (AZ‑900) — Jan 2022
- TensorFlow Developer Specialization — Jan 2021
- AI Foundations for Everyone (IBM) — Dec 2019
- Machine Learning (Andrew Ng) — Jan 2019
Open to roles in ML Engineering, Data Science, and Applied Research.
Let’s talk
Portfolio — projects, code, and demos.