Inbound 8420467398850859716

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

THE AI LAB

ACCELERATOR
By Prudential

Submission Guidelines
Submission Guide
1. Please use this guide to help structure your submission deck.

a. Your submission deck should provide in-depth information of your solution. Include images, diagrams or
charts to illustrate your points.

b. You can add additional information as appendices.

c. Feel free to use your own template design.

d. Refer to the evaluation criteria when drafting your submission to ensure you have adequately
addressed the requirements.

2. Early Submission Deadline: Sunday, 15 Dec 2024, 11.59pm (GMT+8).

3. Final Submission Deadline: Wednesday, 15 Jan 2025 at 11.59pm (GMT+8).

4. Should you have any questions or encounter difficulties in making your submission, please email to
[email protected].

2
Evaluation Criteria
Criterion Weightage Details

● Does the solution offer a unique approach to a problem compared to current


Innovativeness 20%
solutions in the market?

Effectiveness 20% ● How effective is the solution in solving the problem? What is the expected
and Impact impact of the solution on the Prudential business/operation?

● Does the company have the technical and business expertise to execute the
Team and
20% pilot project? What is the company’s track record in developing and
Capability
deploying similar solutions?

● Technical Soundness: Is the AI technology used technically viable and


proven in practice?
Technical
Feasibility and 40%
● Stage of Development: Is the solution at a prototype stage, MVP, or already
Development
in production? Is it clear how the Accelerator programme can provide
assistance to further develop the solution?
3
Submission Deck Template

4
Cover slide
Recommended number of slides in this section: 1
1. Company and submission title

2. One-liner about your proposed AI solution

5
Problem
Recommended number of slides in this section: 1
1. Introduce the problem you are solving. You do not have restate the challenge
statement as it is published. Rephrase and reframe it, if appropriate.

2. Briefly explain your understanding of the problem.

6
Solution
Recommended number of slides in this section: 2
1. Introduce your proposed solution.

2. Describe the solution’s value proposition and competitive advantage(s).

3. Example(s) of implementation (POC/pilot or commercial deployment).

7
Technology
Recommended number of slides in this section: 5 (refer pages 9 to 13)

1. System Diagram: Include a labeled visual.

2. Data Strategy and Management: Text + Visual on data flow.

3. AI Model and Frameworks: Diagram of model structure + key frameworks.

4. Deployment Details: Summary of deployment architecture.

5. Scalability and Security: Bullet points on considerations.

8
Technology - System Diagram
Overview of the System: A top-level diagram showing the major components of the solution and
how they interact. This could include:

● Data Sources: Where the data originates (e.g., CRM, customer touch points, external APIs,
etc.).
● Data Pipeline: Any tools or platforms used for data ingestion, cleaning, transformation, and
storage.
● Processing and Model Deployment: Illustrate how the AI model processes data and produces
output. Include components like machine learning platforms, cloud services (AWS, GCP, or
Azure), and any edge devices, if applicable.
● Integration Points: Identify where the solution integrates with Prudential’s existing systems
(e.g., CRM, ERP, or analytics platforms).

Component Details: A brief explanation of each component, such as databases (e.g., SQL/NoSQL),
cloud infrastructure, microservices, and APIs.
Technology - Data Strategy and Management
Data Types and Sources: Outline the types of data used (e.g., structured, unstructured, customer
data, transaction data, etc.) and the origin of these data sources.

Data Preprocessing and Feature Engineering: Description of steps taken to preprocess data,
including data cleaning, feature selection, and transformation. Highlight any unique techniques used
in preparing the data for model training.

Data Security and Privacy: Key measures for handling sensitive customer information, compliance
with regulations (e.g., GDPR, PDPA, etc.), and security protocols in place (e.g., encryption and
secure access control).

Data Storage and Scalability: Information on data storage solutions, such as cloud databases or
on-premises servers, and strategies for scaling data infrastructure to support larger datasets.
Technology - AI Model and Frameworks
Model Architecture: Describe the model structure, whether it’s a neural network (CNN or RNN),
tree-based model, or another form of machine learning algorithm. Include:
● Model Layers and Parameters: For complex models, brief details on layers, parameters, and
any specialised configurations used.
● Frameworks and Libraries: Mention key frameworks (e.g., TensorFlow, PyTorch, scikit-learn,
etc.) and libraries (e.g., Hugging Face, OpenCV for image processing, etc.) that support model
development.

Training Process and Optimisation: Explain the training pipeline, including:


● Training Data: Quantity, quality, and method of acquiring and labeling data.
● Model Optimisation: Techniques used to improve performance (e.g., hyperparameter tuning,
model pruning, ensembling).
● Testing and Validation: Steps taken to test the model’s accuracy, such as k-fold
cross-validation, and performance metrics used (e.g., accuracy, precision, recall, etc.).
Technology - Deployment Details
Deployment Strategy: Outline the approach for deploying the model, including:
● Cloud vs. Edge Deployment: Whether the model is deployed on cloud servers (e.g., AWS
SageMaker, GCP AI Platform) or at the edge for real-time processing.
● Continuous Integration/Continuous Deployment (CI/CD): Any CI/CD pipelines to automate updates
and deploy new model versions seamlessly.

Model Serving and Monitoring:


● Model Serving: Techniques used to serve predictions, such as REST APIs, gRPC, or message
queues.
● Monitoring and Maintenance: Steps to monitor model performance post-deployment, including drift
detection, retraining strategies, and regular model evaluations.

Integration with existing Ecosystem: Detailed explanation of how the solution interfaces with existing
systems:
● APIs and Middleware: APIs or middleware used to integrate with customer databases, product data,
or analytics.
● Customisations: Any specific customizations to ensure compatibility with existing technology stack,
data formats, and security standards.
Technology - Scalability and Security
Performance, Scalability and Security Considerations
● Performance Metrics: Outline specific technical metrics relevant to the solution, such as
inference time, latency, and computational requirements.
● Scalability: Describe how the architecture can scale to handle larger datasets, increased
requests, or expansion to new features or geographies.
● Security Measures: Emphasis on secure data handling, including compliance with relevant
regulations (GDPR, HIPAA), secure data transmission, and authentication protocols.

Future-Proofing and Adaptability


● Adaptability of the Architecture: Strategies to allow future AI model upgrades or incorporate
new data sources without significant redesign.
● Modular Design: Highlight any modular or microservices-based design choices that enable
easy integration of new functionalities and allow the architecture to evolve with advancements
in AI.
Commercialisation
Recommended number of slides in this section: 2
Growth Potential

1. Elaborate on the technical feasibility of your solution.

2. Estimate the potential market size for your solution.

Scalability

3. Describe (briefly) your business model.

4. Describe how you intend to scale a business around your solution.

14
Details of Proposed Pilot
Recommended number of slides in this section: 3
1. Describe your design of the pilot. If appropriate and possible, include illustrations.

2. Provide a projected timeline (up to 3 months) for the implementation of the pilot.

3. Describe what resources are required to execute the project, including those you
would like to obtain from The AI Lab and Prudential.

15
Company and Team
Recommended number of slides in this section: 2
Company Profile

1. Provide a company introduction.

2. State the relevant track record, investments, and key awards.

Team Profile

1. State the background of key team members involved in this pilot proposal.

2. State their roles, responsibilities, and expertise.

16
Appendices
Recommended number of slides in this section: <5
This is where you add additional information and materials
to elaborate on specific section in the main proposal.

This section is not mandatory.

17

You might also like