How we use GenAI at IBM Software Support

Please share your comments or suggestions on the LinkedIn post. thx

 Empowering Software Support: The Transformative Role of Generative AI

Project Ideas for Fall 2024

 Here are some team project ideas. I expect my students to improve these ideas further.

AI-Driven Endpoint Security for BYOD (Bring Your Own Device) Environments

• Objective: Create a cybersecurity solution that uses AI to monitor and secure devices in a BYOD environment, detecting and mitigating potential risks caused by non-corporate devices accessing the network.
• Novelty: Implement a system that learns from device behavior, network usage, and data access patterns to recommend customized security measures or quarantine devices when suspicious activity is detected.
• Enterprise Use: Ensures security while allowing flexibility for employees using personal devices, especially in hybrid work models.

AI-Driven Identity and Access Management (IAM) System

• Objective: Develop an AI-powered platform that automates identity and access management by analyzing employee roles, data access patterns, and anomalies to recommend optimal access controls.
• Novelty: The system could predict and recommend least-privilege access policies based on AI insights, reducing the risk of excessive access permissions while balancing productivity.
• Enterprise Use: Enhances enterprise security by minimizing access-based risks and improving compliance with regulatory frameworks.

AI-Driven Employee Wellness and Productivity Tracker

• Objective: Build an AI-based platform that tracks employee productivity metrics (e.g., project completion rates, time spent on tasks) and correlates them with wellness data (e.g., stress levels, absenteeism, survey responses).
• Novelty: The platform should suggest personalized interventions (break times, counseling) and productivity boosters using AI models trained on behavioral data.
• Enterprise Use: Improves employee well-being and enhances productivity across teams.

AI-Enabled Enterprise Document Summarizer

• Objective: Create an AI platform that automatically summarizes lengthy enterprise documents (such as meeting notes, reports, and contracts) into concise, actionable insights using NLP techniques.
• Novelty: Integrate the summarizer with enterprise chatbots, allowing employees to query the system for real-time, contextual document summaries and quick answers.
• Enterprise Use: Improves efficiency in handling large volumes of information in daily operations.

Enterprise Expense Fraud Detection System

• Objective: Develop a platform that uses AI and anomaly detection techniques to flag fraudulent or unusual employee expenses in an organization.
• Novelty: The system should learn from employee spending patterns, detect outliers, and also provide insights into common fraud patterns across departments or locations.
• Enterprise Use: Reduces risk and enhances financial integrity.

AI-Driven Talent Acquisition and Retention Tool

• Objective: Develop a tool that uses AI to analyze patterns in employee recruitment and retention. The tool should predict which candidates are likely to succeed and stay longer in the organization based on data from resumes, interviews, and performance reviews.
• Novelty: The platform should continuously improve its model by considering industry trends, skill shortages, and company-specific needs.
• Enterprise Use: Helps HR departments in making better hiring and retention decisions.

AI-Powered Insider Threat Detection System

• Objective: Develop an AI-driven platform that monitors employee activity to detect potential insider threats based on behavior patterns, communication anomalies, and unusual system access.
• Novelty: Include predictive analytics to forecast potential insider threats before they occur, based on changes in employee sentiment or workload (e.g., using sentiment analysis from internal communications).
• Enterprise Use: Prevents data breaches and intellectual property theft caused by internal actors.

Data-Driven Cybersecurity Risk Scoring for Enterprises

• Objective: Develop a platform that calculates a dynamic cybersecurity risk score for an enterprise by analyzing system vulnerabilities, threat intelligence feeds, and employee behaviors.
• Novelty: Integrate external data sources such as social media and dark web activity to provide predictive risk scoring, identifying potential upcoming attacks based on public information leaks or discussions.
• Enterprise Use: Helps organizations prioritize security resources by focusing on high-risk areas and potential threats.

AI-Based Adaptive Firewall System

• Objective: Build an adaptive firewall that uses machine learning to dynamically adjust its rules based on real-time network traffic and emerging threat patterns.
• Novelty: Include the ability for the firewall to autonomously add or remove rules based on continuous monitoring of network health, and also provide insights into why certain rules were adjusted.
• Enterprise Use: Provides more robust and adaptable network defense mechanisms, improving protection against evolving cyber threats.

No-Code API Integration Platform

• Objective: Build a no-code platform where users can easily integrate different enterprise APIs (e.g., CRM, ERP, and HR systems) by dragging and dropping pre-built connectors.
• Novelty: The platform should automatically generate the necessary authentication and API calls, allowing users to visually map data between APIs without needing coding expertise.
• Enterprise Use: Simplifies the process of integrating multiple enterprise systems, making it accessible to non-developers in business teams.

Low-Code Multi-API Security Gateway

• Objective: Build a low-code platform that allows enterprises to secure their API ecosystem by creating custom security rules (authentication, rate limiting, encryption) for multiple APIs through a visual interface.
• Novelty: Include AI-driven security recommendations based on detected vulnerabilities or common security gaps in the enterprise's API infrastructure.
• Enterprise Use: Enhances API security across the organization, allowing security teams to enforce best practices with minimal coding effort.

AI-Assisted No-Code Business Rule Engine

• Objective: Develop a no-code business rule engine that allows users to define and enforce rules for enterprise workflows (e.g., approvals, notifications, data validation) by integrating with APIs and setting conditions through a visual interface.
• Novelty: Incorporate AI to suggest rule optimizations, identify inefficiencies, and predict outcomes based on historical data and trends.
• Enterprise Use: Streamlines decision-making processes in enterprise workflows, reducing errors and speeding up processes by allowing non-developers to create business logic without coding.

No-Code Custom API Builder with AI Code Generation

• Objective: Build a no-code tool that allows users to create their own APIs by defining data structures, business rules, and endpoints through a visual interface.
• Novelty: The platform could use AI to automatically generate backend code (in Python, Node.js, etc.) based on the user's visual inputs, enabling rapid API deployment.
• Enterprise Use: Empowers business users and non-developers to create custom APIs tailored to specific organizational needs without writing a single line of code.

AI-Driven Kubernetes Network Policy Generator

• Objective: Create a platform that generates and enforces network policies for Kubernetes clusters based on traffic patterns and security needs, using AI to analyze and optimize the policies dynamically.
• Novelty: Include features where AI learns from past network behavior to suggest the most efficient and secure policies, improving cluster security without sacrificing performance.
• Enterprise Use: Enhances the security of Kubernetes environments, particularly in large-scale enterprise deployments with complex microservice architectures.

AI-Assisted Kubernetes Deployment Versioning and Rollback

• Objective: Develop a system that manages deployment versioning and rollback strategies in Kubernetes environments, using AI to predict the best time to deploy updates and automatically decide when to roll back based on performance metrics.
• Novelty: The AI model should continuously learn from past deployments to optimize future releases, ensuring that rollback decisions are timely and accurate.
• Enterprise Use: Improves the reliability of application updates in enterprise environments, reducing the risk of downtime or broken features during releases.

Automated AI-Enhanced Penetration Testing Platform

• Objective: Build an AI-driven penetration testing platform that automates the simulation of cyber-attacks to identify system vulnerabilities in real time.
• Novelty: Include adaptive learning features where the AI can tailor penetration tests to specific enterprise environments, making tests more efficient and relevant by learning from the enterprise's past security incidents.
• Enterprise Use: Provides continuous and autonomous security assessments, allowing enterprises to proactively identify and fix vulnerabilities before exploitation.

MS project idea: Using GenAI to convert a monolithic software architecture to a micro services based

Can GenAI help guide developers in converting a monolithic software architecture to a microservices-based one? How?

In a monolithic system, features are tightly coupled. A change in one part of the system can impact the entire application. Scaling a monolithic application often means replicating the application as a whole, which can be inefficient. On the other hand, Microservices allow for independent scaling of each service. If the product catalog experiences high traffic, only that service needs to be scaled. Each microservice can be developed, deployed, and updated independently, improving maintainability and reducing the risk of system-wide failures.

GenAI can play a significant role in helping developers convert a monolithic software 

architecture to a microservices-based one. Here are some ways GenAI can assist:

1. Identifying Monolith Components: GenAI can analyze the codebase and identify potential candidates for individual microservices. It can consider factors like:

* Code complexity

* Functionality

* Data dependencies

* Performance characteristics

2. Service Boundary Detection: GenAI can help detect natural service boundaries within the monolithic application by analyzing code structure, APIs, and data flows.

3. Microservice Candidate Selection: Based on its analysis, GenAI can recommend specific components that are well-suited for being individual microservices, taking into account factors like:

* Complexity

* Business value

* Interdependence with other services

4. Service Interface Definition: GenAI can help define the interfaces between microservices by analyzing API dependencies and identifying potential communication patterns.

5. Data Distribution Planning: GenAI can assist in planning data distribution and storage for individual microservices, considering factors like:

* Data size and complexity

* Storage requirements

* Data replication needs

6. Service Deployment Strategy: GenAI can provide insights on the most suitable deployment strategies for each microservice, taking into account factors like:

* Scalability needs

* High availability requirements

* Network topology

7. Code Refactoring Guidance: GenAI can offer guidance on refactoring code to create independent microservices, suggesting changes like:

* Code structure modification

* API design updates

* Data storage optimization

By leveraging these insights and recommendations, developers can successfully convert their monolithic software architecture to a microservices-based one, ensuring a more scalable, maintainable, and efficient system.

Example:

Let's say we have a monolithic e-commerce platform with a complex codebase. GenAI analyzes the code and identifies potential candidates for individual microservices based on functionality, data dependencies, and performance characteristics. It recommends that the Order Management System, Payment Gateway, and Product Catalog be separate microservices.

GenAI then helps define the interfaces between these microservices by analyzing API dependencies and identifying potential communication patterns. For example, it suggests that the Order Management System communicates with the Payment Gateway via a RESTful API to process payments.

Finally, GenAI provides guidance on refactoring code to create independent microservices, suggesting changes like updating API designs, optimizing data storage, and improving scalability.

Please share your thoughts in the comment below. 

Make Everyone Smile

Hey there! Just wanted to let you know that today is officially National 'Make Everyone Smile' Day! So, consider yourself officially tasked with spreading joy and laughter wherever you go. Oh, and if you happen to fall for any pranks along the way, just remember it's all in good fun! Happy April Fool's Day!

New MS project ideas

 

1. The car tells you what it needs http://news.mit.edu/2017/software-let-your-car-tell-you-what-it-needs-1026
2. Similarly in an Industrial setting, an expensive machine tells you when it needs service
3. Wristbands detect stress signals http://news.mit.edu/2016/empatica-wristband-detects-alerts-seizures-monitors-stress-0309
4. www.tanium.com endpoint management
5. Build a tool like AHA Ideas: https://www.aha.io/pricing?product=Ideas
6. Develop a data pipeline for Climate Intelligence (get inspiration from https://cervest.earth/
7. Morale and culture-building app for remote teams get inspired from https://culture.justdisco.com/
8. cross-selling services for a specific business. Get inspired from https://swogo.com/ Fir example, millions of sellers on eBay or Amazon sell just 1 or 2 items, the service will enable them for cross-sell other products to the same customers and save customer acquisition costs
9. Subscription business creation platforms are emerging, see this for zym/athelitc trainers https://playbookapp.io/

Spring 2021 - MS CMPE272 Team Project Ideas

While COVID taught us a great deal of resilience, patience, empathy and generosity, it also gave us opportunity to think hard and introspect. What's important in our lives. How do we use our technical skills and everything else to solve problems with greater good. 

Data sources: https://usafacts.org/data/

My Grad students always do projects using cool modern technology such as AI and Machine Learning. Past projects are available here:

Spring 2017: https://github.com/SJSU272LabS17

Fall 2017: https://github.com/SJSU272LabF17

Spring 2018: https://github.com/SJSU272LabSP18

Fall 2018: https://github.com/SJSU272LabF18

Spring 2019: https://github.com/SJSU272Spring2019

Fall 2019: https://github.com/SJSUFall2019-CMPE272

Spring 2020: https://github.com/SJSUSpring2020-cmpe272

This semester, they will do even better. Here are some ideas I am sharing with them:

Category: AI in Software Development

Idea#1: Apply ML into Fuzzing to improve software reliability

Use ML to avoid testing every possible input by training on "interesting inputs" making testcases to find more suspect outputs. Read more about Fuzzing here

This area of research is also called as "Big Code" see DARPA announcement 

JSNice is another good implementation of Statistical renaming, type inference and deobfuscation technique on Javascript code. 

Idea#2: New implementation of the Probabilistic programming 

Probabilistic programming or (PPL) is a tool for statistical modeling. The idea is to borrow lessons from the world of programming languages and apply them to the problems of designing and using statistical models. Learn more here. Also see a WebPPL interactive  implementation here. 

Idea#3: Use AI to measure and improve Microservice performance

Microservices architecture is common ways to deploy containerized software in the cloud. However, increasing number of microservices increases management complexity due to distributed architecture. Network latency and load balancing are other challenges. See this use case for inspiration. 

Category: Data Privacy & Governance

Idea#4: Data Privacy in the Decentralized AI architecture

Whether its a race to get COVID vaccine developed or race to open businesses safely, its all about data sharing between government and private entities without compromising privacy of our citizen. There are technologies such as Secured Multi-party computation (sMPC), differential privacy and other  techniques that enable privacy preserved data sharing. Take an example of various shopping Malls in a city sharing data while not disclosing any competitive information. A centralized analysis of federated data will allow decision makers to take safety measures, yet facilitating shopping experience. 

Idea#5: Create domain specific business glossary using NLP

Vertical taxonomies such as an e-commerce taxonomy for retailer, telecom taxonomy for Telecom. Businesses will be able to bring their domain specific private documents and the NLP based AI system will create a well formed taxonomies ingestible into a common catalog such as Google data catalog or Alation Business Glossary

Category: Social Good

Idea#6: Lucy keeps our seniors happy and engaged

#1 problems our seniors (over 80) face today is loneliness and isolation. According to the U.S. Census Bureau 11 million, or 28% of people aged 65 and older, lived alone at the time of the census. As people get older, their likelihood of living alone only increases. Additionally, more and more older adults do not have children, reports the AARP, and that means fewer family members to provide company and care as those adults become seniors.
1. Senior isolation increases the risk of mortality.
2. Feelings of loneliness can negatively affect both physical and mental health.
3. Perceived loneliness contributes to cognitive decline and risk of dementia.
4. Social isolation makes seniors more vulnerable to elder abuse.
5. Loneliness in seniors is a major risk factor for depression.

6. Socially isolated seniors are more pessimistic about the future.

Lucy is a skill developed for Amazon Alexa device echo dot which sits in senior’s room.

A group of student volunteers write personal emails to our seniors . Lucy reads those emails to seniors as they come or at certain time.

Idea#7: Get inspiration from DataKind Projects

https://www.datakind.org/projects

Idea#8: Using Divrsitykids datasets, create a dashboard and use NLG to generate a citizen friendly report

Datasets: https://data.diversitydatakids.org/dataset

Idea#8: Using Divrsitykids datasets, create a dashboard and use NLG to generate a citizen friendly report

Datasets: https://data.diversitydatakids.org/dataset

NLG: https://rosaenlg.org/rosaenlg/1.10.1/index.html

Idea#9: Street vendors support network

Bringing street vendors in an organized group has several advantages:

• Vendors will get much needed financial help and support to bootstrap their business. 
• Govt will have much desired success in getting them on epayments and cashless economy.
• this network will be able to uplift them and help them move into high value quadrant. For example, a street food hawker graduates to become a food truck owner. A vegetable seller eventually becomes a “Fresh cut vegies” supplier to apartments. 
• This network can be connected to micro lenders and other entrepreneurs such as organic farmers.

Category: Cyber Security

Idea#10: Zero Trust security 

• Zero trust vs “trust and verify”
• Hyperlocalization in business will drive next gen endpoint security for data and application access
• Borderless security practices for multi-national companies and workforce
• Remote Work-employee experience

Category: Immigration

Idea#11: Look at the questions to ask and available data sets here ( https://usafacts.org/issues/immigration/ ) and create a dashboard for legal immigrants that can help answer questions such as: 

what are the chances of a H1B visa being denied?

What will be the impact on a certain industry if H1B Visa is eliminated?

What will be the benefit to US economy if US implements time bond citizenship for immigrant workers? 

Category: Crime & Justice

Idea#12 Look at available data sets and create a dashboard for US taxpayers which can answer important questions such as:

• whats the burden on tax payers for prisoners in private prison vs state prison?
• impact of pending court cases on economy and taxpayers?, correlation with specific types of crime?
• correlation between certain types of firearms and crime?

Datasets: https://usafacts.org/data/topics/security-safety/crime-and-justice/

Idea#13: Justice on wheels

Access to Legal services in India is very limited, especially in rural areas. In many cases, long running justice process and expensive legal help deter people even to seek justice. A Justice on wheels app(with lawyers network) that provides mobile (at your doorstep ) service for:

• Affidavit and stamp papers related work
• Will and trust
• Legal rights education 
• Consumer protection related services and others

This “On wheels” idea can be applied to Tutoring, Books and stationary selling.

“Accounting on wheels” can be another opportunity as there are tons of BCOM boys and girls in India looking for work. The company will train them on business and financial accounting

More coming..

Story of a food

 Food is what brings us together, if one thing this pandemic has taught us is that how important it is to cook at home with our family, eat healthy and nutritious food and exercise to boost immunity and stay sane during this time. I love cooking at home, I am a big fan of cooking with healthy grains and vegetables even if they come from unknown territories. Recently I started a blog to share the untold stories of some of the superfoods which I think, folks don't know much about. In this series, I have published stories of Ginger, Beetroots and Quinoa. Please subscribe the Substack blog post so you get weekly stories directly into your Inbox. Let me know your thoughts and any recipes that you want to share. 

https://foodstory.substack.com/