JD

About

James Ding

I'm a data scientist and aspiring AI engineer who builds end-to-end ML systems and AI-powered pipelines. Currently working as an Associate Data Analyst at AirSprint Private Aviation, where I specialize in time-series forecasting, SQL optimization, and data modeling in Snowflake. I hold an MDSA (Master of Data Science and Analytics) from the University of Calgary.

I'm drawn to complex systems — whether that's building autonomous trading bots on AWS Lambda, training XGBoost models that track how oil prices drive Calgary housing, or building RAG systems that predict Supreme Court verdicts. I enjoy the full lifecycle: from messy exploratory analysis to production-ready AI systems.

Recent work:

  • Built Polybot — an autonomous Polymarket trading bot on AWS Lambda using a Haiku→Sonnet Claude cascade with an insider-trading detector
  • Overhauled Calgary Housing Intelligence — 3 XGBoost models, live pressure score, 8-agent geopolitical simulation
  • Built AI Judge — Supreme Court verdict prediction using RAG with Claude and FAISS on 200 landmark cases
  • Working through 93 AI engineering projects covering LLMs, RAG, agents, MCP, and fine-tuning

Education

MDSA — Master of Data Science and Analytics

University of Calgary

Calgary, AB, Canada

BSc — Agricultural and Consumer Economics

University of Illinois Urbana-Champaign

Champaign, IL, USA

Skills

Python SQL TypeScript Time-Series Forecasting XGBoost / scikit-learn Prophet / ARIMA PyTorch EDA & Feature Engineering RAG Pipelines LLM Agents Multi-Agent Systems LangChain LlamaIndex FAISS Ollama CrewAI MCP (Model Context Protocol) AWS Bedrock AWS Lambda API Gateway OpenSearch Serverless AWS CDK S3 Snowflake Amazon QuickSight Tableau Power BI GitHub Actions CI/CD Docker / Compose FastAPI pytest Git & GitHub Astro React Tailwind CSS Business Stakeholder Comms Agile / Scrum
Core ML/Analytics AI/LLMs Cloud & Infra DevOps Web Process

Experience

Associate Data Analyst

AirSprint Private Aviation

May 2025 – Present

Calgary, AB

  • Designed and shipped a production time-series forecasting model (Prophet, with LSTM experiments) for revenue flight-hour planning.
  • Achieved 6% weekly MAPE and 2% monthly MAPE in production; 40% error reduction vs. backtest baseline.
  • Addressed $49M empty-leg cost exposure, identifying $5M–$10M savings opportunity for the operations team.
  • Built Snowflake data pipelines and Amazon QuickSight dashboards consumed by executive stakeholders.
  • Resolved UTC timestamp ingestion bugs and developed volatility-aware smoothing strategies for noisy ops data.

Data Analyst Intern

AirSprint Private Aviation

2024

Calgary, AB

  • Supported EDA and baseline modelling for flight-hour demand, laying the groundwork for the production forecaster.
  • Authored SQL queries in Snowflake for ad-hoc reporting and KPI tracking.
  • Collaborated with operations and finance teams to translate business questions into analytical deliverables.