Lead Data Scientist Forecasting-Remote
-
IT/Tech
Job Description & How to Apply Below
Job Description
HI,
Need Lead Data Scientist with Forecasting-Remote with Travel
Location:
Remote but travel will be expected to Nashville at least 2-3 times a year (paid for by us).
Duration: 12+Months
The customer wants a forecasting specialist with production SRE-like instincts — someone whose first move when a model breaks is to interrogate the data pipeline, not the model architecture. Broad AI/ML generalist experience (LLMs, CV, GenAI) is not a substitute for hands-on, senior-level time-series and Snowflake production experience. This role skews toward "forecasting systems engineer" more than "ML researcher."
Core mandate: someone who can independently design, own, and maintain production forecasting systems — not just build models in a notebook.
Must-Have Technical Depth
1. Time-Series Forecasting Expertise
- Forecast horizon design and rolling / walk-forward forecasts
- Backtesting strategy and temporal validation
- Chronological (not random 80/20) train/test splitting
- Lag feature engineering and seasonality modeling
- Forecast monitoring in production
2. Forecast-Specific Evaluation Metrics
- WMAPE and why it's preferred for multi-site / multi-location forecasting
- Understands MAPE's bias toward over-weighting small-volume locations
- Knows why R² is not a suitable business forecasting metric
- Can reason about business impact of error (e.g., 5% error on 1M visitors vs. 40% error on 500 visitors)
3. Production Ownership Mindset
- When forecasts break, diagnoses upstream causes first: input data validity, feature drift, pipeline changes, SQL transformations, data freshness, model versioning
- Does NOT default to hyperparameter tuning as the first response to a production anomaly
- Thinks like a "production ML owner," not just a "model developer"
4. Feature Leakage Sophistication
- Point-in-time correctness
- Future information contamination
- Feature availability timestamps
- Training/serving skew
- Understands this is distinct from overfitting — PCA / regularization / cross-validation don't address leakage
5. Modern Data Platform Fluency
- Real, senior-level (not "3–4 months learning tables") experience with Snowflake-native ML pipelines
- Snowpark Python
- SQL debugging at scale
- Task orchestration
- Warehouse-scale transformations
6. Deeper Classical ML Knowledge
- LightGBM vs. XGBoost tradeoffs
- Regularization parameters and tree growth strategies
- Categorical feature handling
- Model calibration
- Feature importance interpretation (beyond conceptual familiarity)
Nice-to-Have (Already Validated as Strengths)
- GenAI / LLM / NLP application experience
- General ML fundamentals (classification metrics, precision/recall, F1)
- Computer vision / document intelligence background
(If this job is in fact in your jurisdiction, then you may be using a Proxy or VPN to access this site, and to progress further, you should change your connectivity to another mobile device or PC).
View / Apply for Jobs
Matching My Jurisdiction