Case Studio
Build your own version of work real companies have talked about
Companies write in public about the problems they solve: recommendations, delivery times, fraud, forecasting, defects. Each brief here starts from those pages, the company's own blog, newsroom or reports. You read the case, take a free, open dataset and build your own version for a different market, user or constraint. Every brief comes as a mini-project or a major project.
- 1. Read the caseThe company's own public pages, linked on each brief.
- 2. Make it your ownChange the market, data, users or limits, so the project is yours.
- 3. Build it with open dataFree datasets, never the company's own data, with a mentor's review.
AITeky has no link with these companies, and none of them has reviewed or endorsed these briefs. Company names only say where the idea comes from.
44 cases
- Mini-project Banking & Fintech Case Studio · Razorpay Payment success rate dashboard Find where online payments fail (method, bank, time) and show a merchant where they lose money. Python · Statistics · Data simulation · Power BI or Streamlit
- Major project IT, Software & SaaS Case Studio · Netflix Personalised thumbnail chooser with contextual bandits Choose which image to show each viewer for a title, learning from clicks with a contextual bandit, and test it offline. Python · Contextual bandits · Off-policy evaluation · Computer vision
- Major project IT, Software & SaaS Case Studio · Spotify Playlist continuation and home-shelf ranking Suggest the next songs for a playlist and rank shelves on a home screen, evaluated like a real recommender. Python · Embeddings · Learning to rank · LightGBM
- Major project Manufacturing, Robotics & Industry 4.0 Case Studio · Bosch Production line failure prediction Predict which parts will fail final quality checks from measurements taken along the production line. Python · Big data handling · XGBoost · Imbalanced learning
- Major project Retail, E-commerce & Consumer Goods Case Studio · Flipkart Query understanding and product ranking Understand what a shopper means (category, brand, size, price), then rank products so the right ones come first. Python · NLP · Named entity recognition · Learning to rank
- Major project Banking & Fintech Case Studio · Stripe Real-time payment fraud scoring service Score each payment for fraud in milliseconds with rules plus a model, and give analysts a review queue. Python · Feature engineering · LightGBM · APIs
- Major project Automotive, EV & Aerospace Case Studio · Waymo Road object detection and tracking Detect and track cars, people, bikes and autos in driving video, and test how well it works on Indian roads. Python · Object detection · Tracking · PyTorch
- Mini-project Retail, E-commerce & Consumer Goods Case Studio · Best Buy Search query to product predictor Predict which product a shopper wants from what they typed into an electronics store search box. Python · Text processing · scikit-learn · Search
- Major project IT, Software & SaaS Case Studio · Airbnb Search ranking for a stays marketplace Rank hotel and stay results for each search so guests find and book faster, using learning to rank. Python · Learning to rank · LightGBM · Experiment design
- Mini-project Automotive, EV & Aerospace Case Studio · Waymo Self-driving crash rate comparison Compare crash rates of driverless cars with human drivers fairly, using the right benchmarks. Python · Statistics · Data storytelling
- Mini-project Electronics, Semiconductors & Telecom Case Studio · Intel Semiconductor yield prediction from sensors Predict which wafers will fail from hundreds of process sensor readings, and find the sensors that matter. Python · Feature selection · Classification · Manufacturing analytics
- Mini-project IT, Software & SaaS Case Studio · Spotify Similar-song finder from audio features Find songs that sound alike from their audio features, and group them into moods a listener would recognise. Python · pandas · Clustering · Similarity search
- Major project Banking & Fintech Case Studio · Razorpay Smart payment router with live learning Send each payment to the gateway most likely to succeed, learn from results, and fail over when a gateway degrades. Python · Simulation · Multi-armed bandits · Classification
- Mini-project Retail, E-commerce & Consumer Goods Case Studio · Flipkart Spelling and sound-alike fixer for product search Correct misspelt and sound-alike product searches typed by Indian shoppers, so they still find the product. Python · NLP · Fuzzy matching · Search
- Major project Agriculture & Food Case Studio · John Deere Spot-spraying vision system Detect weeds in field images in real time and decide which nozzles to open, then estimate savings and risks. Python · Semantic segmentation · Edge AI · Simulation
- Major project Manufacturing, Robotics & Industry 4.0 Case Studio · Nestlé Supply chain demand and late-delivery risk model Forecast demand and predict which orders will ship late, then show planners where to act first. Python · Forecasting · Classification · SHAP
- Mini-project Manufacturing, Robotics & Industry 4.0 Case Studio · Bosch Visual defect checker for cast parts Spot defective cast parts from photos so inspectors only look at the doubtful ones. Python · PyTorch · Computer vision · Quality inspection
- Major project Electronics, Semiconductors & Telecom Case Studio · Intel Wafer map failure pattern classifier Recognise failure patterns on wafer maps automatically and feed them into a yield dashboard. Python · Computer vision · PyTorch · Clustering
- Mini-project Agriculture & Food Case Studio · John Deere Weed or crop image classifier Tell weeds from crops in field photos so a sprayer could treat only the weeds. Python · PyTorch · Computer vision · Precision agriculture
- Mini-project Retail, E-commerce & Consumer Goods Case Studio · Walmart Weekly store sales forecast with holidays Forecast weekly sales by store and department, and show how holidays and markdowns change them. Python · Time series · Forecasting · Power BI or Streamlit

