Domain: e-Commerce

Projects, I have worked on:

Project #1: Anomaly detection from the Customer’s Orders amount, using Facebook Prophet Description: Considering business context, we keep close eye on the orders placed by the customers. There can be ‘Global and ‘Contextual outliers’

Project #2: Order count forecast, using Xgboost regressor Description: Orders Delivery team gets benefitted, if we are able to forecast the orders count for the next week or a next month. Using Xgboost regressor, we are able to predict the order count, say for the next week

Project #3: Customer’s tweet analysis for the ‘completed’ orders Description: Tweet's sentiment analysis includes reviews of items catered by large & small e-commerce participants (sellers), and about 350 leading enterprises & brands. Leveraged 10,000+ tweets to develop sentiment analysis model that helped improve sales and marketing strategies

Project #4: Provide multiple text processing solutions to the content writing team Description: Dictionary and grammar support like similar sentence (and Q&A) generation, spelling auto correction, word cloud. In the blogs, language conversion, text summarization and Text-to-speech conversion features were added

Project #5: Text Extraction (OCR) from the Order statement or Payment receipt Description: Provided solution like text extraction from image or pdf invoice file. OCR solution also helps, when support team places an order on behalf of the customer or Catalogue managed for the sellers

Project #6: Product’s labeling and order placement from the image file Description: Customer can share an image and if that item is available in our stock then the order placement workflow gets started. What different items are present in that set of images, recognize and label all those

Project #7: Sales Forecast of multiple Products across different stores Description: On different stores, products have their own sales cycle. Using Apache Spark and fbprophet, we have created multiple ML models to predict products sale

Project #8: Computer Vision:

1. Using "Image classification", predict which class(es) (i.e. items) belong to it. Output is 'score', 'Label", and 'Box'.
2. Using 'Visual question answering (VQA)", ask a question, relevant to the image and get the answer
3. Using 'CLIP', get higher score to a class label, which is more relevant
4. Using 'VisionEncoderDecoderModel ', get reasonable image captioning results
5. YOLO