Case Studies

Here are few case studies that demonstrate our applications of machine learning (ML) and generative AI (Gen AI) in different industries and use cases:

1) Demand Forecasting for Automobiles: Predicting Next Year’s Demand from 5 Years of Data

A leading automobile manufacturer needed to leverage machine learning techniques to forecast the demand for their vehicles in the upcoming year based on the past five years of historical sales data. By utilizing time series forecasting models such as ARIMA (AutoRegressive Integrated Moving Average) and LSTM (Long Short-Term Memory), we successfully predicted demand fluctuations for different vehicle models. The model incorporated factors like seasonal trends, economic conditions, consumer behavior, and past sales data. The forecast helped the manufacturer optimize production schedules, manage inventory levels efficiently, and adjust marketing strategies accordingly, resulting in a significant reduction in overproduction and stockouts, leading to better customer satisfaction and cost savings.

2) Generative AI with RAG (Llama 3.1) for Maintenance Information of Shop Floor Robots

In an advanced manufacturing environment, a company needed to adopt Generative AI powered by Retrieval-Augmented Generation (RAG) using the Llama models to extract critical maintenance information from text PDFs of robot datasheets and machine manuals. By indexing these documents and integrating them with RAG capabilities, we helped their workers to query the system using natural language to retrieve relevant maintenance instructions and troubleshoot issues in real time. This solution reduced downtime significantly by providing quick access to accurate technical details, ensuring that the robots on the shop floor were always maintained optimally. The system improved operational efficiency, enabling faster issue resolution and minimizing costly delays in production.

3) Machine Vision for Quality Inspection of Connecting Rods in Automobile Manufacturing

A leading automobile parts manufacturer asked us to implement a deep learning-based machine vision system for quality inspection of connecting rods using Convolutional Neural Networks (CNN). The CNN model was trained to detect defects such as cuts, dents, scratches, bolt absence, and poor circular shape. The system used high-resolution images of the connecting rods captured on a production line, where the model processed and analyzed each image to identify defects in real-time. The solution automated the quality control process, significantly reducing human error, increasing inspection speed, and ensuring that only high-quality parts were sent for assembly, improving the overall product quality and customer satisfaction.

4) Visual Attribute Prediction for Fashion Categories: T-Shirts and Sarees

We helped a global retail brand to use machine learning to predict visual attributes such as sleeve length, collar type (round neck, polo, or regular), and color from product images in categories like men’s t-shirts, women’s t-shirts, sarees and others. Here, we deployed a convolutional neural network (CNN) for image classification using EfficientNetB4, enabling accurate categorization based on visual features. This AI solution helped streamline the e-commerce product tagging process, reducing manual work and improving search relevance on online platforms. It enhanced customer experience by making it easier for consumers to filter products by their preferred attributes, ultimately driving higher sales conversion rates.

5) Generative AI for Summarization of Legal Documents (Contracts, NDAs, and Other Legal Texts)

A law firm specializing in corporate contracts and legal agreements needed integration of generative AI for automated summarization of lengthy documents, such as contracts and NDAs. Using transformer-based models like Mistral-7B and Llama, the system was developed to identify key clauses, terms, and obligations, generating concise summaries. The AI tool enabled legal professionals to quickly grasp the essence of lengthy legal documents, reducing the time spent on manual review and improving accuracy. This not only streamlined workflow but also enhanced the firm's ability to handle more cases with greater efficiency and accuracy, offering faster turnaround for clients.

These case studies highlight our diverse applications of machine learning and generative AI across industries, demonstrating our ability to solve complex problems, improve operational efficiency, and provide valuable insights.