Data Labs

Work through real business problems with real datasets, and build a portfolio that shows employers what you can do.

OpenAIAIPremiumhard
LLM Evaluation Harness
Build automated graders for two versions of an AI assistant, check them against human labels, and decide which one ships.
- Model evaluation
- LLM as judge
- Automated grading
4–5 hours3 datasets
WayfairSales analyticseasy
E-commerce Sales Analysis
Clean two years of online store orders and turn them into a clear sales review for next year's plan.
- Data cleaning
- pandas
- Time series trends
2–3 hours4 datasets
PelotonBusiness analyticsmedium
Customer Churn Analysis
Find out why subscribers of a fitness app cancel, and give leadership an evidence-backed plan to keep them.
- Churn analysis
- Cohorts and tenure
- Feature exploration
3–4 hours2 datasets
DuolingoMachine learningmedium
Learner Personas with Clustering
Use clustering to discover the kinds of learners using the app, and turn each group into a persona the product team can act on.
- Clustering
- K-means
- Feature scaling
3–4 hours1 dataset
UberOperations analyticsPremiummedium
Driver Supply and Demand Analysis
Measure the ride requests that go unmet for lack of a driver, and recommend where to spend the incentive budget.
- Marketplace operations
- Joins
- Time patterns
3–4 hours4 datasets
InstacartData engineeringmedium
Incremental Order Events Pipeline
Load raw order events one day at a time into clean tables that stay correct with duplicates, late events and a changing schema.
- Data pipelines
- Incremental loading
- Deduplication
3–4 hours4 datasets
RobinhoodMarketing analyticseasy
Marketing Channel Performance
Compare six months of ad spend against signups and paying customers, then recommend where the next budget should go.
- Marketing metrics
- Joins and aggregation
- CAC and ROAS
2–3 hours2 datasets
Amazon Prime VideoMachine learningPremiumhard
Streaming Recommendation System
Build a recommender that ranks ten titles per viewing profile, and beat showing everybody the most popular ones.
- Recommender systems
- Collaborative filtering
- Ranking metrics
4–5 hours3 datasets
AirbnbData engineeringPremiumhard
Bookings Star Schema with Price History
Model bookings as a star schema that keeps listing price history, so past revenue stays correct when hosts change prices.
- Data modeling
- Star schema
- Slowly changing dimensions
4–5 hours3 datasets
EtsyExperimentationhard
Checkout Redesign A/B Test
Analyze a four-week checkout experiment, deal with messy tracking, and decide whether the redesign should launch.
- A/B testing
- Hypothesis testing
- Confidence intervals
3–4 hours2 datasets
DoorDashMachine learninghard
Delivery Time Prediction
Build and evaluate a model that predicts food delivery times, then explain what makes deliveries slow.
- Regression
- Feature engineering
- scikit-learn
4–5 hours2 datasets
WorkdayAIPremiumhard
HR Policy Assistant with RAG
Build a RAG assistant that answers employee questions from policy documents, cites its source and says so when the answer isn't there.
- RAG
- Embeddings
- Vector search
4–5 hours3 datasets
AT&TMachine learningPremiumhard
Retention Offer Uplift Modeling
Measure how much a retention offer changes each customer's chance of churning, then decide who should get the next one.
- Uplift modeling
- Incrementality
- Randomized campaign
4–5 hours3 datasets
CVS HealthHealthcare analyticsmedium
Medication Adherence Analysis
Measure medication adherence from pharmacy refill records, and list the patients an outreach call could still help.
- Pharmacy analytics
- Date logic
- Data cleaning
3–4 hours3 datasets
American AirlinesBusiness analyticsmedium
Flight Delay Analysis
Separate the delay a flight inherits from the delay it creates, and trace the worst delays back to the flights that caused them.
- Root cause analysis
- Window functions
- Time zones
3–4 hours3 datasets
Capital OneMachine learningPremiumhard
Credit Card Default Prediction
Build a model that predicts which credit card applicants will default, decide who to approve, and explain each rejection.
- Classification
- Credit risk
- Data leakage
4–5 hours2 datasets
RazorpayMachine learningPremiumhard
Payment Fraud Detection
Build a classifier that flags fraudulent payments before they're approved, and pick a review threshold based on what fraud actually costs.
- Classification
- Imbalanced data
- Precision and recall
4–5 hours2 datasets
UPSForecastingmedium
Parcel Volume Forecasting
Forecast daily parcel volume for each sort hub through December, and turn the forecast into a staffing plan.
- Time series
- Demand forecasting
- Seasonality
3–4 hours3 datasets
WayfairCustomer analyticsmedium
RFM Customer Segmentation
Score every customer on recency, frequency and spend, build segments, and give the CRM team a plan for each one.
- RFM analysis
- Customer segmentation
- pandas
3–4 hours4 datasets
DoorDashNLPmedium
Support Ticket Triage
Route customer support tickets to the right team from their text, and prove the model beats a keyword baseline.
- Text classification
- NLP
- scikit-learn
3–4 hours2 datasets
StripeNLPhard
Help Center Search
Build a search engine that finds the right help article for questions written in customers' own words, and measure it properly.
- Information retrieval
- Search ranking
- TF-IDF and BM25
3–4 hours3 datasets
WayfairAImedium
Review Insights Extraction with an LLM
Extract structured data from unstructured product reviews with an LLM, and evaluate your prompts against labeled examples.
- LLM prompting
- Structured outputs
- Few-shot prompting
3–4 hours2 datasets