Veritas

VERITAS is a national-level AI and Analytics Case Competition designed to challenge participants to solve complex, real-world business problems through data-driven analysis, artificial intelligence and strategic decision-making.

Participants will work on an integrated business case involving credit risk, fraud detection and financial decision-making. Teams must analyse datasets, identify meaningful patterns and develop an AI-powered solution capable of generating accurate, explainable and actionable insights.

The competition combines data analytics, business problem-solving and AI agent development, offering participants an opportunity to demonstrate both analytical thinking and practical AI implementation skills.

Guidelines

Competition Format

- Number of Problem Statements: 1 integrated business case. - Round Duration: 7 days from the official release of the case. - Registration Deadline: 16 November 2026. - Case Release and Submission Deadline: Will be communicated through Unstop.

The case will focus on an industry-relevant business problem combining financial analytics, credit risk assessment, fraud detection and intelligent decision-making.

Participants will receive a detailed problem statement, supporting business information and a development dataset.

The competition will involve two primary components:

Part A: Data Analytics and Risk Assessment

• Teams will be required to analyse the provided datasets, identify patterns, investigate potential risk indicators and derive meaningful business insights.
• Participants are expected to demonstrate their ability to handle complex datasets, address data-quality challenges and develop suitable analytical methodologies.

Part B: AI-Powered Business Solution

• Teams must design an AI-powered assistant or analytical solution capable of applying their findings to business scenarios.
• The solution should demonstrate the ability to interpret financial and operational information, identify risks, make recommendations and provide explainable outcomes.
• Participants may use platforms such as Gemini, ChatGPT, Python-based frameworks or other suitable AI and analytics tools.

Development and Hidden Evaluation Datasets

- To ensure objective assessment, the competition will use two separate datasets:

Development Dataset:

• Provided to all registered teams along with the case statement.
• Used for exploratory data analysis, solution development and validation.
• May contain financial records, business relationships, transaction patterns and risk indicators.

Hidden Evaluation Dataset:

• Maintained confidentially by the organising team.
• Used to assess submissions against previously unseen business scenarios.
• Designed to evaluate generalisation, analytical accuracy, consistency and decision-making capabilities.

Solutions will be evaluated on their performance against unseen data, rather than solely on their performance on the provided development dataset.

Each team must submit the following before the specified deadline:

- Solution Presentation (maximum 3 slides): Explaining the analytical approach, key insights, proposed solution and business impact. - AI Agent / Model Configuration: A reproducible version of the developed AI assistant, including instructions, model configuration, code or equivalent implementation files, as applicable. - Analytical Outputs: Results in the prescribed format, including risk scores, predictions, recommendations or other outputs specified in the case. - Supporting Files: Relevant scripts, prompt configurations, model files or documentation necessary to reproduce the solution.

Detailed submission templates and technical requirements will be shared along with the case statement.

Rules and Regulations

- All participants must register through Unstop before the registration deadline. - A participant may be associated with only one team. Multiple registrations by the same participant across different teams are not permitted. - Teams may consist of 1–4 members. Inter-college and inter-specialisation participation is permitted. - Each team must submit only one final solution. Multiple submissions or updates after the deadline will not be considered unless expressly permitted by the organisers. - All submissions must be original work developed by the participating team for the competition. - Participants are permitted to use generative AI tools, machine learning frameworks, programming languages and publicly accessible analytical resources. However, they must clearly disclose the tools and methodologies used. - The use of paid AI subscriptions is not mandatory. Teams will not receive additional marks merely for using premium AI models or tools. - Teams must ensure that their submitted solution is reproducible and compatible with the evaluation requirements specified by the organisers. - The hidden evaluation dataset must remain confidential. Any attempt to access, manipulate or obtain unauthorised information about the hidden evaluation data will result in disqualification. - Participants must not fabricate analytical results, manipulate submitted outputs or misrepresent the capabilities of their solutions. - Plagiarism, unauthorised collaboration between teams or submission of another team's work will lead to disqualification. - Participants must adhere to the prescribed submission format, file requirements and deadlines. - The organising committee may request additional information, clarifications or demonstrations to verify the authenticity and functionality of a submitted solution. - Late, incomplete, inaccessible or corrupted submissions may be rejected at the discretion of the organising committee. - Participants must use the provided synthetic datasets responsibly and must not submit personal, confidential or sensitive third-party information. - Teams are expected to maintain academic integrity and professional conduct throughout the competition. - The organising committee reserves the right to modify the competition schedule, guidelines or evaluation process if necessary. Any material changes will be communicated to participants through official channels. - The decisions of the organising committee and evaluation panel will be final and binding.

Important Dates

-

Last Date for Registration: 16 November 2026

- Case Release: To be announced - Final Submission Deadline: 7 days after case release, as specified on Unstop - Results Announcement: To be announced