Winnipeg · B.Sc. Statistics & Mathematics
I build AI agents that run 24/7. 34 scheduled bots + 17 on-demand agents on a shared zero-token LLM client - 8 model tiers, semantic cache, supervisor, circuit breakers. The whole fleet has cost $0.004 so far. Themed as Jurassic Park, obviously.
| Project | What it is | Links |
|---|---|---|
| 🦖 JURASSIC PARK | The whole fleet in one repo: 34 scheduled bots + 17 on-demand agents, shared zero-token LLM client (8 model tiers), semantic cache, supervisor, circuit breakers, and the live dashboard - species catalog, park console terminal, budget ledger, DNA-bank cache. Nightly refresh. | site · repo |
| ⛅ Weather Arbitrage | Weather-derivative trading bot: NOAA degree-day deviations → arbitrage windows → API orders | repo |
| Project | Tools | What it shows |
|---|---|---|
| 🦠 COVID-19 Global Dashboard | Python · Streamlit · Plotly | Global tracker - cases, deaths, choropleth maps, time-series |
| 🏥 Medicare Fraud Detector | SQL · Streamlit | Cross-country phantom billing detection, risk scoring |
| 🧠 QLoRA Fine-Tuning | Unsloth · QLoRA | 8GB VRAM instruction tuning, zero-cost seed data |
| 🐧 Palmer Penguins | R · ggplot2 | Linear models, ANOVA, ANCOVA |
| 📞 Telco Churn | R · Python | Logistic regression, ROC/AUC |
| 🚲 Bike-Share Demand | R · Python | Count regression, GLMs |
| R · Python | ARIMA/SARIMA forecasting |
- Scaling the fleet: new crews, better caching (semantic L2, prefix L3)
- Keeping it all at $0 - free tiers, local models, cache-first
- Weather arbitrage: more markets, tighter detection
- Data projects with SQL + Python + dashboards
34 bots · 4 crews · 8 species · one very tired GPU