img
img

Shell-Edunet Skills4Future AICTE Internship

  1. Home
  2. Internship
Internship 2024-25 Internship 2025-26
Internship 1 Internship 2

Internship Type: Virtual
Internship Title: Edunet Foundation | Shell | Artificial Intelligence with Green Technology | 4-weeks Virtual Internship
Internship Description:

Dive into the world of Artificial Intelligence with Green Technology and unlock the door to a future filled with innovation and opportunity!

Join the Shell-Edunet Skills4Future AICTE Internship! This is your chance to immerse yourself in hands-on learning of essential technical skills for success. Shell-Edunet Skills4Future AICTE Internship is designed to bridge the employability gap by equipping students with essential technical skills in both Artificial Intelligence (AI) and Green Skills. This certificate-linked program seeks to empower the learners to thrive in the rapidly evolving skill ecosystem, fostering their ability to build successful careers in the dynamic technology sector. Through applying the knowledge of Artificial Intelligence in an efficient way along with the Green Skills to solve the sustainability goals of the society.

Industry experts will mentor throughout the internship. You'll have the opportunity to develop project prototypes to tackle real-world challenges by using your preferred technology track. Work in a student team under your mentor's guidance, you will work in a student team to identify solutions to problems using technology. Selected students will also have the chance to showcase their developed project prototypes at a regional showcase event attended by industry leaders.

About Shell:

Shell is a global energy and petrochemical company operating in over 70 countries, with a workforce of approximately 103,000 employees. The company's goal is to meet current energy demands while fostering sustainability for the future. Leveraging diverse portfolio and talented team, the company drives innovation and facilitates a balanced energy transition. The stakeholders include customers, investors, employees, partners, communities, governments, and regulators. Upholding core values of safety, honesty, integrity, and respect, the company strives to deliver reliable energy solutions while minimizing environmental impact and contributing to social progress.

About Edunet:

Edunet Foundation (EF) was founded in 2015. Edunet promotes youth innovation, tinkering, and helps young people to prepare for industry 4.0 jobs. Edunet has a national footprint of training 300,000+ students. It works with regulators, state technical universities, engineering colleges, and high schools throughout India to enhance the career prospects of the beneficiaries.

Keywords:

AI, Power BI, MI, Data Analytics, Green Skilling, Python Programming, Artificial Intelligence, Computer Vision, Deep Learning, Generative AI, Dashboard Programming, Microsoft Excel, Sustainability

Locations: Pan India
No. of interns: 3000
Amount of stipend per month: ZERO
Qualification: Engineering – 2nd, 3rd & 4th Year Students
Specialization:
Engineering- Computer Science, IT, Electronics and Communication, Electrical engineering, Mechatronics, Data Science

Link: https://internship.aicte-india.org

Perks:
  • Personalized mentorship sessions and collaborative group learning.
  • Opportunities to expedite learning through project-based internships.
  • A holistic learning experience provided by industry experts through knowledge-sharing sessions.
  • Showcase your skills by creating prototypes to solve real-world challenges.
  • Earn certifications from AICTE, Edunet and Industry Partners, boosting your confidence and value to potential future employers.
  • Opportunity to present your project prototypes to a panel of industry experts at a regional showcase event.

Terms of Engagement: 4-Weeks (15th July to 16th August 2025 )

Last date to apply: 30th June 2025

Eligibility Criteria:
  • Age: 17+
  • Pursuing degree in computer science, IT, electronics, mechatronics, and related fields.
  • Students must be able to commit required hours for program in addition to regular academics
  • Students must have basic computer operating and programming skills, as relevant
  • Any exposure to programming is preferred but not mandatory
  • Students should have access to computer/laptop with internet connection, either owned OR through institution

Note: The enrolment of students in the 4-weeks Skills4Future virtual internship is subject to the discretion of the team responsible for the operationalization of the Internship at Edunet Foundation.

Indicative timelines for the internship:
Event Timeline
Onset of registration 03-06-2025
Closing applications for internship registrations 30-06-2025
Orientation of Internship 11-07-2025
Commencement of internship 15-07-2025
Offer letter disbursement for internees 16-07-2025
End of internship 16-08-2025
Awarding certificates 20-08-2025

Advance Machine learning and Artificial Intelligence Project

  1. Water Quality Prediction
  2. Greenhouse Gas Emission Prediction
  3. E-Waste Generation Classification
  4. Garbage Classification
  5. Carbon Emissions Prediction

Weekly Completion Tasks

Weekly Completion Tasks

Week 1:

Project Planning and Data Preparation.

  • Define the business problem and set project objectives.
  • Gather relevant datasets and explore potential data sources.
  • Clean and preprocess data by handling missing values, outliers, and encoding.
  • Perform exploratory data analysis (EDA) to understand data patterns.
  • Split data into training, validation, and test sets.

Week 1: 

  • Define the problem and project objectives.
  • Collect and clean the dataset.
  • Perform EDA to understand the data.
  • Split data into training, validation, and test sets

Submission Details:  

Expected content: Student should create github repository and they should upload their jupyter notebook File (.ipynb) on Github repository and share link on week1 submission page

File format: GitHub Repository link where your partial project is uploaded

Project Submission Link - On LMS

Skills4future.in Via GitHub link

  

Week 2: Model Selection and Building

  • Research and choose appropriate models for the task.
  • Implement a baseline model and evaluate its performance.
  • Train various machine learning models (e.g., Random Forest, SVM, Deep Learning).
  • Conduct feature engineering to improve model performance.
  • Apply cross-validation for more reliable model evaluation.

Week 2:

  • Research and choose appropriate models.
  • Implement a baseline model and evaluate it.
  • Train different models and tune hyperparameters.
  • Perform feature engineering for improvement.
  • Use cross-validation to check model reliability 

Submission Details: 

Expected content: Expected content: The student must show the partial output with the help of Jupyter Notebook, saving, sharing the project link where it is uploaded on GitHub link

File format: .ipynb file, .py file

Project Submission Link -

Skills4future.in Via GitHub link

  

Week 3: Model Evaluation and Optimization.

· Evaluate models using metrics like accuracy, precision, recall, or RMSE.

· Fine-tune models through hyper parameter optimization and regularization.

· Perform error analysis to address under fitting or over fitting issues.

· Implement ensemble methods like bagging or boosting if needed.

· Use model interpretation techniques to explain predictions.

· Testing and Iteration 

  • Formatting 
  • Submit the Project 

Week 3:

· Evaluate models using relevant metrics.

· Tune hyper parameters for better performance.

· Perform error analysis to refine the model.

· Implement ensemble techniques for boosting performance.

· Interpret model output.

Submission Details: 

Expected content: The student must show the output with the help of Jupyter Notebook, saving, sharing the projects, etc. And also create PPT for project.

File format: .ipynb file, .py file, PPT

Project Submission Link - On LMS

Skills4future.in Via GitHub link

  

Week 4: 

Mock Presentation & Final Presentations 

Week 4: Students should present the project PPT to Experts
Water Quality Prediction

About the Project

This project aims to build a forecasting tool that helps estimate river water pollutant levels. The predictions can support environmental agencies in taking timely action and developing sustainable water quality policies.

The project involves:
  • Loading and cleaning real-world pollutant data
  • Feature extraction and transformation
  • Predictive modelling using Random Forest
  • Model evaluation and real-time deployment using a web interface

Learning Objectives

To predict water safety or potability by forecasting multiple pollutant levels using historical station-wise data and interpreting the predicted values to estimate an overall pollution level.

Dataset Acquisition and Preparation:
  • Gather and clean the dataset to ensure quality and consistency.
  • Understand how various pollutants affect overall water safety.

Exploratory Data Analysis (EDA):
  • Analyze distributions and trends of water quality indicators.
  • Visualize spatial and temporal pollutant variations.
  • Identify outliers and correlation among pollutants.

Feature Engineering:
  • Extract useful features like year, month, and station location.
  • Encode categorical data (e.g., station ID).
  • Consider upstream/downstream dynamics for modeling.

Machine Learning Model Development:
  • Implement models like Random Forest Regressor wrapped in MultiOutputRegressor for multiple output prediction.
  • Random Forest was selected for its robustness, performance, and ease of interpretability.
  • Tune hyperparameters to optimize model accuracy.

Model Evaluation:
  • Use metrics like Mean Squared Error (MSE) and R² Score.
  • Compare actual and predicted values for pollutants like PM10, NO₂, O₃, SO₂.

Data Source Link: https://www.kaggle.com/datasets/vbmokin/wq-southern-bug-river-01052021
Greenhouse Gas Emission Prediction

About the Project

This project focuses on predicting greenhouse gas emissions related to U.S. industry supply chains using machine learning. By training models on publicly available datasets from authoritative government sources, the project aims to equip users with insights into supply chain emissions and promote data-driven sustainability practices. The workflow includes data ingestion, preprocessing, EDA, model building, evaluation, and interpretation.

Learning Objectives

The objectives of this project are to:

Dataset Acquisition and Preparation:
  • Load and clean the dataset from data.gov.
  • Handle missing values, incorrect entries, and ensure data consistency.

Exploratory Data Analysis:
  • Analyze the structure, distribution, and relationships within the dataset.
  • Visualize trends in emission factors across industries and commodities.

Feature Engineering:
  • Encode categorical features like Substance, Unit, Commodity, and Industry.
  • Incorporate important numerical and categorical DQ (Data Quality) metrics.
  • Normalize or standardize data as needed using scaling techniques.

Machine Learning Model Development:
  • Build and evaluate multiple regression models: Linear Regression, Decision Trees, Random Forests, Gradient Boosting, etc.
  • Compare performance metrics such as MAE, RMSE, and R² score.

Data Source Link: https://catalog.data.gov/dataset/supply-chain-greenhouse-gas-emission-factors-for-us-industries-and-commodities
E-Waste Generation Classification

About the Project

This project aims to develop an automated e-waste classification system using advanced deep learning techniques, specifically leveraging EfficientNetV2B0. By accurately categorizing electronic waste from images, the model will streamline sorting processes for efficient recycling. This initiative seeks to mitigate environmental and health risks associated with improper disposal of e-waste. The ultimate goal is to create a reliable tool that contributes significantly to sustainable waste management and environmental protection.

Learning Objectives

The objectives of this project are to:

Dataset Acquisition and Preparation:
  • To meticulously prepare the E-Waste Image Dataset, ensuring data integrity, consistency, and standardized input for model processing.
  • This includes comprehending the folder-based classification structure to accurately infer class labels.

Exploratory Data Analysis:
  • To visually inspect and analyze image characteristics, class distribution, and potential imbalances within the dataset.
  • This step aids in identifying challenges like intra-class variability or inter-class similarities.

Feature Selection and Engineering:
  • To understand how transfer learning implicitly leverages powerful, hierarchical pre-trained features from extensive datasets like ImageNet.
  • This recognizes the advantages of using EfficientNetV2B0 as an efficient feature extractor, conserving training time and resources.

Deep Learning and Transfer Learning Model Development:
  • To comprehend transfer learning's benefits (reduced training time, feature exploitation, enhanced performance) and EfficientNetV2B0's operational principles (Fused MBConv blocks, progressive learning).
  • This involves implementing and adapting the EfficientNetV2B0 architecture for e-waste classification by incorporating custom layers.

Model Evaluation:
  • To train the developed model on the e-waste image dataset, monitoring progression with accuracy and loss metrics.
  • This includes rigorously evaluating performance on an unseen test dataset using classification metrics like confusion matrixes and classification reports, and analyzing class-specific performance.

Data Source Link: https://www.kaggle.com/datasets/akshat103/e-waste-image-dataset
Garbage Classification

About the Project

This project leverages EfficientNetV2B2 and transfer learning for garbage image classification. It automates the classification of waste materials, such as cardboard, glass, metal, paper, plastic, and trash. The result is a deployable model that enhances waste processing efficiency and encourages proper recycling through automation.

Learning Objectives

The objectives of this project are to:

Dataset Acquisition and Preparation:
  • To meticulously prepare the Garbage Image Dataset, ensuring data integrity, consistency, and standardized input for model processing.
  • This includes loading the dataset from a folder-based structure using “image_dataset_from_directory”, resizing all images to a uniform size (124×124), and applying batch processing for efficient training.

Exploratory Data Analysis:
  • To visually inspect and analyse image characteristics, class distribution, and any potential imbalance across the garbage categories.
  • This step helps identify challenges such as overlapping visual features between classes (e.g., cardboard vs. paper), and guides decisions for model training and evaluation.

Feature Engineering:
  • To understand how transfer learning automatically leverages robust, hierarchical visual features from a model pre-trained on large-scale datasets like ImageNet.
  • This includes appreciating how EfficientNetV2B2 acts as a powerful feature extractor without the need for manual feature design, thereby conserving both training time and computational resources.

Machine Learning Model Development:
  • To implement and fine-tune the EfficientNetV2B2 architecture for the garbage classification task.
  • This involves freezing base layers to retain generalized visual patterns and adapting upper layers to specialize in distinguishing between garbage categories such as metal, plastic, and trash.

Model Evaluation:
  • To train the model using the prepared training and validation sets, tracking learning progression through accuracy and loss metrics.
  • Post-training, the model is rigorously evaluated on a deterministic test dataset using confusion matrices and classification reports to ensure strong performance across all six garbage categories.

Data Source Link: https://www.kaggle.com/datasets/farzadnekouei/trash-type-image-dataset
Carbon Emissions Prediction

About the Project

This project involves predicting CO₂ emissions per capita across 90+ countries using Random Forest Regression. It combines historical climate, population, and economic data from the World Bank Climate Change Data repository. The forecasting is done for a period of 20 years beyond the latest available data and deployed using Streamlit to provide an interactive interface. Users can select countries, view forecasted emissions, inspect key trends, and download future predictions.

Learning Objectives

The objectives of this project are to:

Dataset Acquisition and Preparation:
  • Acquire, clean, and transform climate and socio-economic datasets.
  • Understand the relationship between economic/environmental features and CO₂ emissions per capita.

Exploratory Data Analysis:
  • Visualize emissions data and key indicator distributions over time.
  • Identify patterns, trends, and correlations across different countries and timeframes.

Feature Engineering:
  • Derive new features like growth rates using Compound Annual Growth Rate (CAGR).
  • Transform categorical and time-series data into predictive variables.
  • Select features with high correlation to the target variable (co2_per_cap).

Machine Learning Model Development:
  • Train and evaluate models including Random Forest Regressor with hyperparameter tuning.
  • Use recursive feature elimination and cross-validation to select optimal features.
  • Project emissions 20 years into the future using forecasted features.

Model Evaluation:
  • A Random Forest Regressor with hyperparameter tuning and feature elimination yielded the best performance.
  • Training and testing splits were validated using k-fold cross-validation.
  • The model performed well on unseen country data with minimal overfitting.
  • Random Forest Regressor achieved the highest R-squared score on both the training (0.986) and test data (0.983) compared to other algorithms.

Data Source Link: https://datacatalog.worldbank.org/dataset/climate-change-data

Python

  • Introduction to Python: Python, created by Guido van Rossum, is a versatile programming language widely used for web development, data analysis, artificial intelligence, and more.
  • Setting up your Python environment: Choose an Integrated Development Environment (IDE) like Jupyter or VSCode and install libraries using package managers like pip to set up your Python environment efficiently.
  • Data types and variables: Python supports various data types such as numbers, strings, lists, and dictionaries, providing flexibility for diverse programming needs.
  • Operators and expressions: Python offers a range of operators, including arithmetic, comparison, and logical operators, allowing concise expression of complex operations.
  • Conditional statements: Employ conditional statements like if, Elif, and else to execute specific code blocks based on different conditions in your Python programs.
  • Looping constructs: Utilize looping constructs, such as for and while loops, to iterate through data structures or execute a set of instructions repeatedly.
  • Functions: Define functions to encapsulate reusable code, pass arguments, and return values, promoting code modularity and readability in Python.
  • Basic data structures: Python's fundamental data structures, including lists, tuples, and dictionaries, empower efficient storage and manipulation of data in various formats.
  • Data manipulation: Master data manipulation techniques like indexing, slicing, and iterating to extract and transform data effectively in Python.
  • Working with files: Learn file handling in Python for tasks like reading, writing, and processing data from external files.
  • Introduction to modules and libraries: Leverage powerful Python libraries like NumPy for numerical computing and Pandas for data manipulation and analysis to enhance your coding capabilities.
  • Resources:

Power BI

What is Power BI (Business Intelligence)?

Imagine a toolbox that helps you turn a jumble of raw data, from spreadsheets to cloud databases, into clear, visually stunning insights. That's Microsoft Power BI in a nutshell! It's a suite of software and services that lets you connect to various data sources, clean and organize the information, and then bring it to life with interactive charts, graphs, and maps. Think of it as a powerful storyteller for your data, helping you uncover hidden trends, track progress toward goals, and make informed decisions.

Useful Links for Self-Study:

Exploratory Data Analysis (EDA)

  • Introduction to EDA: Exploratory Data Analysis (EDA) involves systematically analyzing and visualizing data to discover patterns, anomalies, and insights, playing a crucial role in understanding the underlying structure of the data.
  • Importing and loading Data: Data can be imported into Python using various formats such as CSV, Excel, or SQL, providing a foundation for EDA and subsequent analysis.
  • Data cleaning and preprocessing: Cleaning and preprocessing steps, including handling missing values, outliers, and inconsistencies, are essential for ensuring the accuracy and reliability of the data.
  • Descriptive statistics: Descriptive statistics, encompassing measures of central tendency and dispersion, offer a summary of the main characteristics of the dataset.
  • Data visualization: Visualizations like histograms, boxplots, and scatter plots provide a powerful means to explore data distributions, relationships, and outliers, enhancing the interpretability of the dataset.
  • Identifying patterns and relationships: EDA enables the identification of patterns and relationships within the data, helping to uncover hidden insights and guide subsequent analysis.
  • Univariate and bivariate analysis: Univariate analysis focuses on individual variables, while bivariate analysis explores relationships between pairs of variables, offering a comprehensive understanding of the dataset's structure.
  • Feature engineering: Feature engineering involves creating new features from existing data, and enhancing the dataset with additional information to improve the performance of machine learning models.
  • Hypothesis generation: EDA findings often lead to hypothesis generation, fostering a deeper understanding of the data and guiding further research questions or analytical approaches.
  • Resources:

Data Visualization

  • Principles of data visualization: Effective data visualizations prioritize clarity, ensuring that the intended message is easily understandable, and accuracy, representing data truthfully and without distortion.
  • Choosing the right chart: Select appropriate chart types, such as bar charts, pie charts, line charts, or maps, based on the nature of your data and the insights you aim to convey.
  • Matplotlib and Seaborn libraries: Matplotlib and Seaborn are powerful Python libraries for creating both simple and advanced visualizations, providing flexibility and customization options.
  • Customizing visuals: Customize visual elements, including colors, labels, axes, and titles, to enhance the overall aesthetics and effectiveness of your data visualizations.
  • Interactive visualizations: Utilize libraries like Plotly and Bokeh to create interactive visualizations, allowing users to engage with and explore data dynamically.
  • Data storytelling: Data storytelling involves using visuals as a narrative tool to communicate insights effectively, making data more accessible and compelling for a broader audience.
  • Best practices for presenting visualizations: When presenting data visualizations, adhere to best practices such as providing context, focusing on key insights, and ensuring clarity to effectively convey the intended message.
  • Resources: