Next-Gen Streamlining of Practical Examinations of Programming

Courses with AI-Enhanced Evaluation System

Jayesh Sarwade

, Afrin Shikalgar

, Chaitanya Modak

, Nikita Nagwade

Dnyaneshwari Devmankar

and Ganesh Arbad

Department of Information Technology, JSPM’S Rajarshi Shahu College of Engineering, Pune, Maharashtra, India

Keywords: Programming Assessment, Multi-Language Compilers, AI-based Proctoring, Machine Learning Evaluation,

Decision Trees, Support Vector Machines, MERN Stack, Browser Lock API, Dialogflow Chatbot,

Convolutional Neural Networks (CNN), Exam Integrity, Cheating Prevention, Scikit-Learn, OpenCV,

TensorFlow.

Abstract: The final assessment of any course must reflect its goals and contents. An important goal of our foundational

programming course is that the students learn a systematic approach for the development of computer

programs. Since the programming process itself is a critical learning outcome, it becomes essential to

incorporate it into assessments. However, traditional assessment methods e.g. oral exams, written tests, or

multiple-choice questions are not well-suited for evaluating the process of programming effectively.

Additionally, in educational institutes, teachers often use physical chits to distribute problem statements

among students, then students perform them on computers in college labs, which is time-consuming. If

essential compilers or development tools are missing on college computers, students resort to online

compilers, increasing the risk of internet misuse for copying solutions. There is also a possibility of students

using USB drives to share unauthorized code, compromising exam integrity. Therefore, there is a growing

need for a fair and standardized evaluation process that accurately assesses students based on their coding

abilities, eliminating the risk of cheating or unfair advantages. To address these challenges, this paper

proposes a comprehensive software solution to modernize practical exams. The system automates problem

statement allocation and integrates multi-language compilers through APIs like JDoodle and HackerRank. It

ensures exam integrity by enforcing a full -screen mode using Browser Lock APIs, disabling copy-paste

functionality, and adding watermarking for security. The solution includes AI-based chatbots for guidance,

Neural Networks for face detection. For automated and fair evaluation, machine learning models developed

with scikit-learn are employed, using algorithms such as Decision Trees and Support Vector Machines. The

platform is built on the MERN stack, comprising MongoDB, Express.js, React, and Node.js, to ensure a

robust, scalable, and efficient examination process.

1 INTRODUCTION

Practical examinations remain a vital approach to

evaluate students' competencies in subjects like

programming and data structures. However,

traditional methods often lead to inefficiencies,

security vulnerabilities, and instances of academic

dishonesty. To resolve these issues, this document

introduces a comprehensive software solution

designed to modernize practical evaluations and

deliver measurable results. The proposed system

addresses key concerns of standard techniques by

enhancing efficiency, safety, and fairness. With a

user-friendly interface, educators can easily design,

schedule, and manage examinations, specifying

crucial details such as date, time, course, division, and

problem descriptions. The ability to upload multiple

problem statements simultaneously using Excel

sheets significantly reduces preparation time, thereby

streamlining the entire exam setup process. Students

benefit from a robust platform to undertake

assessments, review completed evaluations, and

manage their profiles. Additional features, like multi-

language compilers and an AI-based chatbot, aid in

246

Sarwade, J., Shikalgar, A., Modak, C., Nagwade, N., Devmankar, D. and Arbad, G.

Next-Gen Streamlining of Practical Examinations of Programming Courses with AI-Enhanced Evaluation System.

DOI: 10.5220/0013590100004664

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Futuristic Technology (INCOFT 2025) - Volume 2, pages 246-252

ISBN: 978-989-758-763-4

comprehending problem statements and provide

instant assistance, reducing misunderstandings. The

software's full-screen exam functionality and

automatic question generation features foster fairness

and lessen opportunities for cheating, thereby

minimizing dishonest practices. By automating

examination procedures and integrating advanced

technologies, this software solution not only saves

time but also enhances the overall test experience. It

advocates for eco-friendly practices by eliminating

paper-based assessments, mitigating grading

logistical challenges, and promoting academic

integrity. Implementing this software could

revolutionize practical evaluations in educational

environments, enabling students to effectively

demonstrate their skills in a secure and equitable

digital setting. This system sets a new standard for

practical assessments, indicating significant

improvements in exam efficiency, security, and

learning outcomes. To ensure the system's reliability

and continuous enhancement, AI-driven evaluation

methodology is employed to analyze performance

metrics, identify improvement areas, and provide

actionable insights for refining exam content and

processes.

2 ENHANCEMENTS OVER

TRADITIONAL PRACTICAL

EXAMINATION APPROACHES

2.1 Automation and Efficiency

There is always a lot of instructional work involved in

getting ready for the test, including scheduling the

time, assigning questions, and creating study notes.

Instructors must handle submissions, submit

questions, and keep track of student registration,

which can be a time-consuming and possibly

disruptive procedure. Using this book can be a slow,

ineffective, and poorly managed procedure. With the

help of the interactive user interface, teachers can

quickly design tests by entering information like the

date, time, group, and explanation questions. Manual

data entry is eliminated by automatically managing

student records and uploading problem descriptions

in batches. Teachers can concentrate on teaching

instead of handling the burden since automation

streamlines the testing process, lowers the possibility

of human error, and simplifies workload management.

2.2 Dynamic Problem Allocation

Conventional Method: Issue statements are typically

provided manually in conventional situations, often

with the aid of static templates or predefined lists.

Because of this regularity, students could foresee

questions or even share information with others in

advance. Additionally, because some students might

be given easier or more identifiable problems, manual

assignment raises the risk of prejudice. To address

these problems, the system employs a dynamic

problem allocation technique. Students in the same

batch are given problems at random to ensure that no

two students receive the same set of questions. This

randomization reduces the likelihood of cheating

because pupils are unable to discuss answers or

prepare for specific problems. Given that every

student is assigned a distinct task.

2.3 Real-Time Monitoring

Conventional Method: Invigilators have historically

kept an eye on student behavior during practical tests.

This approach is not infallible, though, as one

invigilator would not be able to supervise every

student efficiently, particularly in bigger groups or

online environments. Due of human oversight’s

unpredictability and susceptibility to diversions,

dishonest operations may go undetected. The

suggested software makes use of technology to

provide reliable, non-intrusive real-time monitoring.

Throughout the test, features like facial detection

confirm the student's identification at regular

intervals, guaranteeing that the registered student

stays at the workstation. Voice surveillance also picks

up on audio irregularities that can point to

communication with unapproved parties. Secure

testing is ensured by this real-time, AI- enhanced

surveillance, which also serves as a disincentive

against dishonest behaviour and preserves exam

integrity.

2.4 AI-Assisted Evaluation

Conventional Method: Practical test evaluation is

usually done by hand, which can be a time-consuming

and labour-intensive procedure. Because various

evaluators may interpret and assess student

submissions differently, manual grading carries the

danger of prejudice, inconsistency, and human error.

This lack of uniformity may result in unfair evaluation

results. The suggested system incorporates automatic

evaluation driven by AI, which reliably and

efficiently grades student work. The AI is able to

Next-Gen Streamlining of Practical Examinations of Programming Courses with AI-Enhanced Evaluation System

247

assess code according to preset standards, including

efficiency, functionality, and conformity to best

practices. Subjectivity is removed, guaranteeing that

every student is evaluated using the same criteria.

Teachers are relieved of the tiresome chore of manual

assessment since the automated evaluation drastically

cuts down on the amount of time needed for grading,

enabling findings to be processed and distributed

more quickly.

2.5 Multi-Language Support for

Programming

Conventional Method: Students are often restricted to

using specific programming languages that are easy

for examiners to evaluate, which may make it harder

for the students to demonstrate their abilities,

especially if they speak languages that are not

supported. Students who are more comfortable

speaking one language but are forced to use another

language may also find it challenging. Proposed

System: The software's multilingual capability allows

students to write their code in the language they are

most comfortable with, allowing them to capitalize on

their strengths and resulting in a more accurate

assessment of their abilities. The platform supports

multiple programming languages, accommodating a

wider range of technical competencies and different

learning requirements.

2.6 Intelligent Assistance During

Exams

Conventional Method: Students may have trouble

comprehending problem statements on traditional

tests, which could impair their performance. aid is

frequently scarce, and invigilators may not be able to

aid with complicated queries without inadvertently

giving away too much. Proposed System: Students

who might require assistance comprehending problem

statements can receive real-time support from the

integrated AI chatbot. It can provide clarifications or

explanations without giving away answers, enabling

pupils to move forward with assurance.

3 ALGORITHM AND SOFTWARE

The development of online examination software

involved the integration of advanced technologies to

guarantee the security, scalability, and efficiency of

the system. Python was selected as the primary

programming language for its adaptability and

extensive library support, enabling seamless backend

development, management of intricate logic, data

processing, and integration with other system

components. The incorporation of Natural Language

Processing (NLP) aimed to enhance the precision of

automated assessments. NLP algorithms were used to

assess textual responses, scrutinize content, grammar,

and structure, automating the grading of essay-type

questions and identifying plagiarism by cross-

referencing submissions with known sources to

uphold academic integrity (Nayak, Surabhi, et al. ,

2022)

NLP was utilized to automatically evaluate

student answers, emphasizing content relevance,

grammatical accuracy, and structural quality,

ensuring consistent and impartial grading of essay-

style questions. Additionally, NLP was employed for

plagiarism detection by comparing responses with a

repository of previously known sources, ensuring

academic honesty and reducing manual intervention

in grading, thereby streamlining the process for

educators. (Prathyusha, Premasindhu, et al. , 2021)

A robust infrastructure for server-side operations

was established using the Django web framework,

offering built-in features for user authentication,

session management, and security. This framework

efficiently managed workflows such as exam

creation, delivery, and submission. MySQL was

chosen as the database system due to its capability to

handle large volumes of structured data, such as exam

questions, student records, and submissions, ensuring

rapid and efficient data retrieval during the exam

process. (Kumar, Choubeya, et al. , 2020)

To facilitate flexible deployment, the software

was containerized using Docker, ensuring consistent

deployment across different environments by

packaging both the software and its dependencies,

minimizing configuration issues and enhancing

scalability. Git was utilized for version control,

enabling developers to track the development process,

manage collaboration, and maintain a history of

changes, thereby facilitating efficient project

management and debugging. (Brkic, Mekterovic, et

al. , 2020)

4 METHODOLOGY

4.1 Project Scope and Requirements

Defining Core Modules and Features: The project

began with a solid understanding of the basic elements

that comprise the practical test software. These

comprised teacher and student modules, test design,

INCOFT 2025 - International Conference on Futuristic Technology

248

scheduling, coding interfaces, issue statement

presentation, incorporating an AI-powered chatbot,

evaluation, and results output. All requirements were

accurately documented because each of these

components was well defined. Analysis of Skills and

Traits: Identifying the specific attributes needed for

each module was an essential initial step.

4.2 Conduct Market Research

Examining Current EdTech Solutions: In order to

comprehend the current state of educational

technology, extensive market research was carried

out, with a special emphasis on platforms that provide

coding tests, AI-assisted learning, and experiential

learning settings. Finding opportunities and gaps

where the suggested method may provide clear

benefits was the aim.

Analysis of Competitors: The strengths,

shortcomings, and areas of differentiation of

competitor products were evaluated. For example,

current platforms may provide coding interfaces

without support for multi-language compilers or

automated grading yet lack strong security features.

The creation of distinctive characteristics that would

distinguish the program was guided by this analysis.

4.3 Develop a Project Plan

Thorough Planning: A thorough project plan was

made that included all of the tasks, due dates, and

resource allocations. Every stage of the project

lifecycle, from original design and development to

deployment and maintenance, was addressed in the

plan. Better project management and budgeting were

made possible by the inclusion of time and cost

estimates.

Establishing Milestones and Deliverables: For

every stage of development, specific deliverables

were set.

4.4 Design the Software Architecture

Scalable and flexible Design: Future updates and

expansions are made possible by the software

architecture's flexible design. Every module,

including the user dashboards, exam management,

and problem statement repository, was created as an

independent part that could be changed without

impacting the system as a whole.

Structural Planning: The architecture contained

information about the processing, retrieval, and

storage of data.

4.5 Discuss the Implementation

The implementation of the system resulted in a highly

efficient and secure platform for conducting practical

exams, addressing the inefficiencies of traditional

methods. By leveraging the MERN stack, the system

provided a robust backend for seamless data

processing and a responsive frontend for user

interaction. Faculty could effortlessly create,

schedule, and manage exams with bulk uploads of

problem statements, saving significant time. Students

benefited from an intuitive interface, enabling

distraction-free coding with secure features like full-

screen mode and disabled copy-paste functionality.

AI-powered evaluation ensured unbiased and accurate

grading, while automated result generation

significantly reduced manual effort and errors. The

modular architecture enhanced scalability and

allowed the system to handle concurrent exams

efficiently, achieving faster load times and higher user

satisfaction. Feedback-driven refinements improved

usability and reliability, ensuring a future-proof

solution that streamlined practical examinations and

upheld academic integrity.

4.6 Present Use Cases and Scenarios

Examples of Practical Uses for Instructors and

Learners: The system was designed with real-world

scenarios in mind. For example, educators can quickly

create assessments, schedule tests, and monitor

ongoing sessions by inserting issue statements as

Excel files into their dashboards. After checking in

and utilizing a unique code to access their evaluations,

students were able to focus on coding tasks in a

secure, distraction-free setting.

Common processes and interactions within the

system: Provide instances of standard practices such

as student registration, exam scheduling, problem

statement dissemination, and automatic grading.

These scenarios ensured that every interaction was

straightforward and seamless by focusing on the user

experience.

4.7 Propose Evaluation Metrics

Creating Success Criteria: Assessment tools were

created to determine the software's effectiveness in

several domains. These metrics included user

satisfaction, dependability, usability, and system

performance. Quantitative measures including

Next-Gen Streamlining of Practical Examinations of Programming Courses with AI-Enhanced Evaluation System

249

average load times, error rates, and successful exam

completions were integrated with user feedback.

The following key performance indicators (KPIs)

were prioritized: scalability (e.g., the ability to handle

multiple concurrent exams), usability (e.g., ease of

navigation), and security (e.g., successful detection of

illegal activities). Other KPIs included the speed and

accuracy of AI-assisted tests and the ability to support

many languages without performance degradation.

User satisfaction and continuous improvement:

Teachers and students took part in surveys and

feedback sessions to find out more about user

satisfaction

5 LITERATURE SURVEY

Table 1: Literature Survey.

Sr.

Publisher

Yea

Title Technologies Benefits Drawbacks/

Limitations

1. IRJET,

June 6,

2022

Online Examination

System Using AI

Machine Learning,

Pattern Matching,

Naive algorithm.

Linguistic

Analysis

orith

Malpractice can be

detected easily.

Only applicable for

theory questions

2. eLifePress,

2022

Online Exam Portal

System Using ML

algorithm

Machine Learning,

Python

Used by students who are

studying for examinations

to practice and track their

ress.

No API that meets the

requirements.

3. IJIRT,

2021

An Examination

System Automation

Using NLP

PYTHON, NLP Immediate Feedback for

errors, Provided solutions

can be accessed.

Applicable only for

multiple

choice type of

questions.

4. ICAISC,

2020

A Study on Web

based Online

Examination System

JS programming

language, Ajax

technics, Mysql

The system's effectiveness

as they can rapidly select

the finest reply given,

minimizing time spent on

each address.

Applicable only for

MCQ questions.

5. IEEE,

Nov2020

Automatic Analysis

and Evaluation of

Student Source

Codes

Machine learning,

Roslyn API

Smooth review process,

automatic assessment of

submitted task

Only C# coding

language is available

no others

IEEE,

April 28,

2020

Building

Comprehensive

Automated

Programming

Assessment System

Python

programming

language, Django

web framework,

MySQL, Docker,

Git version control

stem

Improved scalability,

Reduction in grading time,

Increased consistency in

grading,

Enhanced efficiency in

assessing programming

assi

nments

Limited adaptability,

Lack of real-time

feedback capabilities,

Static nature of the

system,

Absence of adaptive

learnin

features.

7. JETIR,

April,

2015

A Survey on

Integrated Compiler

for Online

Examination Syste

MEAN stack,

JVM, Graph

mining

Conduct both subject

quizzes and lab exams

online

Focus only on

conducting exams

without providing an

evaluation mechanis

8. IEEE,

Dec, 2006

Assessing Process

and a Product- A

Practical Lab Exam

for an Introductory

Programming Course

Programming

languages, web

development

frameworks, and

educational

assessment tools

Hands-on learning,

Real-world assessment,

Collaborative problem-

solving, Instructor

feedback, Understanding

est

ractices

Time-consuming

grading,

Limited scalability,

Resource-intensive

setup, Subjective

evaluation

INCOFT 2025 - International Conference on Futuristic Technology

250

6 PROPOSED MODEL

Figure 1: Proposed Model.

6.1 Instructor Dashboard

Instructors must initially register on the platform with

a username and password, and they can subsequently

serves as a centralized platform to manage

assignments, personal information, and course

administration. Instructors can view and manage their

personal information. Course administration tasks,

such as adding, viewing, and managing courses, are

facilitated by the instructors. Additionally, instructors

can create exams by specifying details such as the

date, time, course, division, and batch. They can also

upload problem statements individually or in bulk via

an Excel sheet, and these problem statements are

randomly assigned to students on the platform.

6.2 Student Dashboard

Students must also register on the platform with

credentials, and they can log in to the system using

these credentials. Just like faculty members, students

have access to a homepage and profile section. On the

dashboard, students can start their exams according to

the assigned problem statement. They can execute

their problem statement on a multi-language

compiler, with an AI-powered chatbot available to

assist in understanding the issued statement. After

completing the main execution, students need to

answer five AI-generated questions related to the

problem statement they worked on. After submitting

the exam, it is automatically evaluated by AI, ensuring

accurate grading, and a detailed exam report is

generated.

7 CONCLUSION

The concept presented here presents an intriguing

opportunity to fundamentally change how educational

institutions assess students: the creation of practical

test software. This explanation aims to give a solid

basis for the implementation phase by defining the

project's characteristics, needs, and scope. The

proposed initiative is to improve the process of

creating assessments, boost student participation, and

speed the administration of tests. It is designed to

satisfy the needs of educators as well as learners. The

proposal makes use of AI chatbots, issue statements,

test design, scheduling, coding interfaces, and

outcome evaluation in an effort to provide a

comprehensive answer to current issues in education.

8 FUTURE SCOPE

Advanced AI models and offline functionality

together enhance the security and accessibility of

online exams. AI models leverage computer vision

and machine learning to detect suspicious behaviors,

such as unusual eye movements or excessive

keyboard activity, ensuring a fair and secure

environment with real-time alerts or automatic

flagging. Simultaneously, offline functionality allows

students to download exam materials and work

without internet access, with automatic

synchronization of completed work once connectivity

is restored. This combination ensures exams are both

inclusive and secure, catering to diverse regions and

Next-Gen Streamlining of Practical Examinations of Programming Courses with AI-Enhanced Evaluation System

251

technological challenges while maintaining academic

integrity.

REFERENCES

“Mr. Vasanth Nayak, Shreyas, Surabhi T, Dhrithi, Divyaraj

K M” have implemented, “Online Examination System

using AI” in “International Research Journal of

Engineering and Technology. (IRJET)”, Vol:09, 2022.

"Siddhant Chaurasia, Govid Suryakant Shinde, Shubham

Rasal, Sudhanshu Singh” have implemented, "Online

Exam Portal System Using Machine Learning

Algorithm. (eLifePress)", Vol:3, 2022.

“T.S.S.K. Prathyusha, K. Premasindhu, T. Ganga Bhavani,

V. stuthimahima” have implemented, “Examination

system Automation using natural Language Processing.

(IJIRT)”, Vol:08, 2021.

“Avinash Kumar, Anjali Choubeya, Ayush Ranjan Behra c,

Anil Raj Kisku d, Asha Rabidase, and Beas Bhadraf”

have implemented, “A study on the Web-based Online

Examination System”, in the International Conference

on Recent Trends in AI, IOT, Smart Cities &

Applications.

“Adam Pinter, Sandor Szenasi”, have implemented, “The

Automatic Analysis and Evaluation of Student Source

Codes”, in IEEE International Conference on

Computational Intelligence and Informatics,2020.

"Ljiljana Brkic, Igor Mekterovic, Boris Milasinovic, and

Mirta Baranovic", have implemented, "Building

Comprehensive Automated Programming Assessment

System", in University of Zagreb, 10000 Zagreb,

Croatia,2020.

“Chetan Kumar, Chinmayee Madhukar”, have

implemented, “A Survey on Integrated Compiler for

Online Examination System”, Vol: 2, 2015.

“Jens Bennedsen, and Michael E. Caspersen”, have

implemented, “Assessing Process and a Product - A

Practical Lab Exam for an Introductory Programming

Course”, at Aarhus University.

Jayesh Mohanrao Sarwade , Harsh Mathur “Scaling Of

Medical Disease Data Classification Based On A

Hybrid Model Using Feature Optimization” Webology,

Volume 18, No. 6, 2021 Pages 3681-3696.

“Ashwini S, Bhuvan Kumar M S, Chandra Sai D; Sathya

V”, have implemented “ProctorXam - Online Exam

Proctoring Tool”, in “2024 International Conference on

Electronics, Computing, Communication and Control

Technology(ICECCC)”,

doi:10.1109/ICECCC61767.2024.10593954

J. Sarwade, S. Shetty, A. Bhavsar, M. Mergu and A.

Talekar, "Line Following Robot Using Image

Processing," 2019 3rd International Conference

on Computing Methodologies and Communication

(ICCMC), Erode, India, 2019, pp. 1174-1179, doi:

10.1109/ICCMC.2019.8819826.

“Adiy Tweissi, Wael Al Etaiwi, Dalia Al Eisawi”, have

implemented, “The Accuracy of AI-Based Automatic

Proctoring in Online Exams”, at Princess Sumaya

University for Technology (PSUT), Jordan.

Jayesh Mohanrao Sarwade, Harsh Mathur “Performance

analysis of symptoms classification of disease using

machine learning algorithms” Turkish Journal of

Computer and Mathematics Education (TURCOMAT),

Volume 11 Issue 3 Pages 2024-2032.

Shudhodhan Bokefode, Jayesh Sarwade, Kishor Sakure,

Sandeep Bankar, Surekha Janrao, Rohini Patil “Using a

Clustering Algorithm and a Transform Function,

Identify Forged Images”, Vol 5, Issue 1, Pages781789

,DOI:10.47857/irjms.2024.v05i01.0299

Sarwade J. M., Sandip Bankar, Janrao, S., Sakure, K., Patil,

R., Bokefode, S., & Kulal, N. (2023). A Novel

Approach for Iris Recognition System Using Genetic

Algorithm. Journal of Artificial Intelligence and

Technology, 4(1),9–17.

https://doi.org/10.37965/jait.2023.0434

“Shilpa Manohar Satre, Shankar M. Patil, Vishal

Molawade, Tushar Mane, Tanmay Gawand, Aniket

Mishra”, have implemented, “Online Exam Proctoring

System Based on Artificial Intelligence”, in 2023

International Conference on Signal Processing,

Computation, Electronics, Power and

Telecommunication (IConSCEPT).

“Aditya Nigam, Rhitvik Pasricha, Tarishi Singh,

Prathamesh Churi”, have implemented, “A Systematic

Review on AI-based Proctoring Systems: Past, Present

and Future”.

INCOFT 2025 - International Conference on Futuristic Technology

252