HackRx 4.

PROBLEM STATEMENT I

1. GENERATIVE AI | HEALTHBOT
Have a medical query?
Google gives too many results and getting the precise answer is time consuming.
ChatGPT gives me precise answers, but we are not sure of authenticity as we do not
know the sources.
To address this issue, develop a Chatbot platform that seamlessly integrates with
reputed knowledgebase sources, including website like:
1. Mayo Clinic [https://mayoclinic.org/]
2. Web MD [https://www.webmd.com/]
3. Authenticated databases
This module will link to the corona self-checker for initial identification of
risk and to the Corona test booking module for escalation.

2. EXPECTED OUTCOMES
1. An end-to-end Chatbot platform
2. Configurable options to enable or disable sources
3. Engaging end customer's chatbot interface with accurate and relevant
responses.
4. BONUS: Scope for continuous improvement based on users' feedback
5. BONUS: Ability to integrate over predefined set of documents

3. JUDGING CRITERIA
1. Accuracy and Relevance: Are the responses relevant to user queries?
2. Configurability: How well does the platform allow users to configure
and enable/disable sources on the fly?
3. UI/UX
4. Technical Implementation: How well is the backend data pipeline
designed and implemented? Is it optimized or is it brute force?
5. Cost to maintain and deploy for large number of end users
6. BONUS: Scope for continuous improvement based on users' feedback
 HackRx 4.0

PROBLEM STATEMENT II

1. GENERATIVE AI | HEALTHGENIE
Empowering Personalized Notifications and Dynamic Pages for a Health-centric
Mobile or Web App

2. PROBLEM STATEMENT
Join us in a groundbreaking expedition, where innovation takes center stage.
Develop a visionary solution that harnesses the cutting-edge potential of
generative AI to create captivating content for notifications and build
personalized pages for mobile or web application. At Bajaj, we cherish fresh
perspectives and eagerly await your out-of-the-box ideas. By tapping into a wealth
of health data stored in a centralized repository (health-mart), your solution will
empower users with timely, relevant, and personalized information, driving their
engagement and inspiring positive health behaviors.

3. KEY REQUIREMENTS
1. AI-Driven Content Generation for Notifications:
Unleash the potential of generative AI to craft a system that
comprehends user profiles, health data, preferences, and contextual
information, producing immersive content for notifications. Seamlessly
deliver timely updates, enticing reminders, personalized
recommendations, and informative snippets that deeply resonate with
users' health goals, interests, and activities.
2. Tailored Page Generation
Shape an extraordinary system that dynamically generates
personalized pages within the mobile or web app, breathing life into
user profiles and interactions. Enchant users with an unparalleled
experience by offering tailor-made health recommendations,
enlightening articles, interactive tools, captivating visualizations,
progress tracking features, and other bespoke content that sets new
standards for personalization.
 4. EVALUATION CRITERIA

1. Unparalleled Personalization:
Assess the level of personalization achieved through AI -powered content generation for notifications and
personalized pages. This involves utilizing user profiles, preferences, and contextual information to create
the best possible experience.
2. Engaging User Experience:
Evaluate the appeal and ability of the content to drive user engagement. The generated content for
nudges and personalization on pages should be enticing and encourage users to interact more.

NOTE:
You can either utilize and provide solution by extending data provided below or construct data in different
segments by your choice. But ensure data should revolve around health insight of user.

EXAMPLE DATA:
Interactions on Web/App (user browsed following
Demographics sections) Doctor Visits Lab Visit Results
Customer
ID AGE CITY ARTICLES SEARCHES SYMPTOMS Visit 1 Visit2 DIABETES LIPID THYROID
Immunity, General
1111222201 36 MUMBAI Ayurveda Fatigue acne Physician - - HIGH -
Ayurveda,
1111222202 39 MUMBAI Nutrition - toothache dentist - - HIGH BORDER
General
1111222203 37 MUMBAI Nutrition Pediatrition - Pediatrition Physician - - -
Children
1111222204 39 MUMBAI - Doctor dentist - - - BORDER HIGH
VERY
1111222205 38 MUMBAI - - - dentist - - BORDER HIGH

1111222206 39 PUNE Nutrition - fever - - - - -

Old Age General
1111222207 37 PUNE - Body Pain fever Physician - VERY HIGH HIGH -

1111222208 35 PUNE Nutrition - fever - - - - -

Heart Health, General
1111222209 40 PUNE Nutrition Weakness - Physician - - - -

1111222210 40 PUNE - - ent - - HIGH - -

General VERY
1111222211 33 BANGALORE - Diabetes gastroenterologist Physician diabetologist BORDER HIGH -
Heart Health,
1111222212 44 BANGALORE Nutrition Gynacologist Gynacologist - - - - -
Heart Health, Cardiac
1111222213 39 BANGALORE Nutrition - - Surgeon - BORDER BORDER -

1111222214 39 BANGALORE - - Muscle Pain - - BORDER - HIGH

Ayurveda, VERY
1111222215 42 BANGALORE Nutrition Gynacologist Gynacologist - - - - HIGH

1111222216 46 HYDERABAD - - - dentist diabetologist - - -

Immunity, VERY
1111222217 45 HYDERABAD Ayurveda Diabetes fever - - - HIGH -
General
1111222218 34 HYDERABAD Nutrition Dentist toothache Physician dentist BORDER - HIGH
Heart Health, Children VERY
1111222219 41 HYDERABAD Nutrition Doctor - Pediatrition - - HIGH -
General
1111222220 35 HYDERABAD - - anxiety Physician - - - -

1111222221 33 DELHI - Pediatrition - - - HIGH - -

Old Age General
1111222222 46 DELHI Nutrition Body Pain dermitologist Physician dermitologist VERY HIGH BORDER -
VERY
1111222223 36 DELHI - Fatigue - - - BORDER HIGH BORDER
General VERY
1111222224 45 DELHI Ayurveda - - Physician - - - HIGH
Immunity,
1111222225 39 DELHI Ayurveda Gynacologist dermitologist - - - BORDER BORDER
EXAMPLE USE CASE BASIS ABOVE DATA:

Scenario:
User A, identified as ID 1111222201, is a 36-year-old individual from Pune. He
recently browsed an article related to Ayurveda and conducted a search for
symptoms associated with fever. User A utilized the BFHL mobile app to book a
doctor's appointment and subsequently visited a lab in the following days.

Objective:
The goal is to personalize User A's notifications in order to enhance their activity
within the app. Additionally, constructing nudges and creating a journey around the
article will indirectly encourage User A to make more bookings and explore the
various plans offered by the company.

We anticipate that you will address the issue by generating additional

use cases. We're excited to see your groundbreaking solution.
 HackRx 4.0

PROBLEM STATEMENT III

1. REALTIME MARKET DATA CHART IMAGE GENERATION

Bajaj Financial Securities Limited (BFSL) is a 100% subsidiary of Bajaj Finance
Limited (BFL). BFSL is heavily leveraging technology to create a unique and real-
time trading experience for its customers and partners.
BFSL trade applications (Mobile and Web) has a Watchlist feature to track scrips
changing prices during market hours.
Currently we show only +/- percent(%) change in prices. We want to include chart
images along with current prices against each script.

Current Proposed
2. REQUIRED SOLUTION
Join us in a groundbreaking expedition, where innovation takes center stage.
• We want a solution to create a snapshot of the charts in realtime, the
duration for the charts will be T-3 months in timeframe.
• The solution will generate chart snapshots in PNG format in realtime and
store in a repository.
• Whenever a request is received say, "ADANIENT" the system should be able
to fetch the last 3 months charts snapshot in PNG format directly from the
repository.
• The system should be designed to accept minimum 1 and maximum 50
scrips in a single request.
• If 3 scrip’s charts image is requested, the system should return only 1
image with 3 charts images merged as below.

• The above output should be in a sequence of the requested scrips.

• As the price changes the image in the repository should get refreshed
automatically.

3. DATASET
• We have attached a link to download last 3 months 50 scrips prices in an
excel sheet.
• The solution should be scalable enough to integrate a live feed from
exchange (NSE/BSE), instead of a static dataset mentioned above.
 HackRx 4.0

PROBLEM STATEMENT IV

SUGGEST ME A STOCK
Bajaj Financial Securities Limited (BFSL) is a 100% subsidiary of Bajaj Finance
Limited (BFL). BFSL is heavily leveraging technology to create a unique and real-
time trading experience for its customers and partners.
In BFSL trade applications (Mobile and Web) we want our clients to get a unique
experience and hence, we would like a feature, ‘Suggest me a stock’.
The recommendations must be based on client’s preference (Risk Profile, Holding,
Watchlist and Stock Search Patterns, etc.), current market trends, macro and micro
economic events, etc.

REQUIRED SOLUTION
Join us in a groundbreaking expedition, where innovation takes center stage.
• Whenever a client logs in or in an interval of 3 days whichever is higher a
suggested stock list should pop up.
• Each client must be given a maximum of 20 scrips as suggested stocks

DATASET
• We have attached a link to download holdings and watchlist data of 50
clients in an excel sheet.
• The solution should be scalable enough to integrate a live streaming data
from various sources instead of a static dataset mentioned above.

DATA HANDLING
Client Stocks in holding
C1 TCS, WIPRO
C2 BATA
C3 TCS
Recommended stocks for client C3: WIPRO

REASON
• TCS and WIPRO belongs to IT sector
• Client C1 is holding TCS as well as WIPRO hence, client C3 may want to buy
WIPRO since he already holds TCS
 HackRx 4.0

PROBLEM STATEMENT V

GAMIFICATION OF PERSONAL FINANCE MANAGEMENT

Empowering Personalized Notifications and Dynamic Pages for a Health-centric
Mobile or Web App

PROBLEM STATEMENT
Gamify the personal finance management for a user to track his financial habits
and well-being and plug Bajaj markets financial offerings where ever a need is
present, ensure a social engagement element within the app for users to interact
with each other. Raw data to be provided, use any additional alternate data from
messages and E-Mails for analysis. Develop a gamification module that creates
personalized challenges, milestones, or achievements based on the spending
analysis. These gamified experiences should incentivize users to improve their
financial habits and save more effectively

EVALUATION CRITERION
• Engagement Value
• Alternate data collection source
• Novelty of idea
• Privacy and security
• User experience
• Incentivization System
• Future scalability
• Integrations
 HackRx 4.0

PROBLEM STATEMENT VI

GAMIFYING DRIVING BEHAVIOR AND FITNESS: FROM

DATA TO DISCOUNTS
Empowering Personalized Notifications and Dynamic Pages for a Health-centric
Mobile or Web App

PROBLEM STATEMENT
Using the accelerometer and GPS data from a mobile phone, collect various metrics
related to driving behavior, such as acceleration, braking intensity, cornering
speed, and overall smoothness of the drive. These metrics can be processed and
fed into a scoring model that assigns a driving score to each user.
Fitness app data can be utilized to gather information about a user's fitness profile.
This data may include daily steps, distance covered, calories burned, heart rate,
sleep patterns, and other relevant health-related metrics. By analyzing this
information, you can create a scoring model to assess the user's overall health and
fitness level.
Incentivize users based on score and offer various rewards and benefits like
discount on motor insurance premium and health insurance premium and also
provide Value Added Services with discounts

EVALUATION CRITERION
• Accuracy of scoring model
• Relevance of Metrics used
• Data Collection

• Privacy and security

• User experience
• Incentivization System
• Scalability and integration
 HackRx 4.0

PROBLEM STATEMENT VII

CREATE BUSINESS FUNNEL

Harnessing social media and public platforms as a strategy to source qualified leads
for business.

BRIEF
You are to help prepare a model, using which social media, external sources and
other online digital avenues can be put to use to collate relevant details for
prospects (who may need protection / investment for their life goals), qualify on
certain gate criteria and provide some relevant insight which may assist in
segmentize the prospect on various parameter. It would help org underwrite /
validation each prospect if the model could gather name, contact details, job title,
context, city / state and other detailed attributes.

DETAILED DESCRIPTION
The model / bot / User Interface should be written to crawl thru various targeted
websites, social media and other part of online digital avenues and systematically
collect various relevant attribute of a prospect lead. For this purpose, part of
crawling may require authorised credential and remaining can be available as
anonymous. It should provide dashboard on completeness and maturity of various
segment of prospects. All attribute of gathered prospect should be secured in
database and be available for instant retrieval. It would be important if the model
could provide additional insight about maturity and segmentation of curated
prospect.

EXAMPLE
Anonymous prospects as well as loyal and connected prospects that are follower of
BALIC, its official tags and its senior members can be the initial phase of collation
to business funnel creation.
 HackRx 4.0

PROBLEM STATEMENT VIII

ONBOARD BUSINESS SMOOTHLY

Harnessing Simplification of customer & partner journey – (Partner
Onboarding Journey, Lead Maturing Journey, Insurance Application Journey etc).

BRIEF
BALIC business manages a most of its processes using digital applications and
BoTs. All significant processes are executed through various user journeys. Most
often these journeys involve authentication, payment, document upload, document
verification and parsing, manual verification etc. Hence most of the journeys are
lengthy , complicated and lack required intelligence. It is expected that candidates
may pick up one or many journeys and work on simplifying it to that extent
possible. In this process candidate may make some important assumptions and use
external services digitally.

DETAILED DESCRIPTION
The workflow and user interface should be developed with clear agenda that the
journey should be intelligent, engaging and simple. It should include decent degree
of intelligent nudges, using internal as well as external services to do maximum
prefill (for existing as well as new to BALIC customer/Partner). One may choose
one or many journeys from list of below option:
(a) Partner Onboarding
(b) Lead Nurturing
(c) Journey for New Application – new to BALIC Customer
(d) Journey for New Application – existing BALIC Customer
(e) Sales Team Productivity Tracking Journey
User personal can be BALIC employee, BALIC partner and / or BALIC Customer.

EXAMPLE
Simplification and intelligent journey for New Partner onboarding or new
Application journeys in a way that users may have only a fewer click and pages.
 HackRx 4.0

PROBLEM STATEMENT IX

PROBLEM STATEMENT: FORGERY OF DOCUMENTS

Empowering Personalized Notifications and Dynamic Pages for a Health-centric
Mobile or Web App

BRIEF
Forgery/tampered documents submitted during insurance claims poses a significant
challenge for insurance companies and can result in financial losses, fraudulent
payouts, and compromised trust in the system.

SOLUTION OPPORTUNITIES
The challenge here is to identify the forged/tampered documents submitted
during claims by to altering, counterfeiting and creating fake/fabricated
documents by various editing tools.

SOLUTION REQUIRED
Our aim is to get a robust system that can effectively identify forged
documents submitted during insurance claims across all lines o f business
(LOBs)

SUCCESS CRITERIA

Accuracy: The solution should achieve a high level of accuracy in detecting

forged documents, minimizing false positives and false negatives.
Comprehensive Coverage: It should be able to identify various types of
forged documents, including but not limited to altered documents, counterfeit
documents, and fabricated documents.
Efficiency: It process documents efficiently, providing quick results to
minimize delays in claim processing while maintaining accuracy.
Scalability: It should be scalable to handle large volumes of claims and adapt
to new types of forgery techniques as they emerge.
 HackRx 4.0

PROBLEM STATEMENT X

PROBLEM STATEMENT: THIRD PARTY LEGAL CLAIMS

BRIEF
Preparation of Written statements, objections, Interlocutory applications, written
notes of arguments with inputs on specific defences take lot of time causing delay
and thus jeopardizing our defence before the court.
Indian and foreign Judgements, citations, Research, case studies, articles on
specific law and on available evidence in cases takes a lot of time for reading and
manual search.
Proper and effective Repudiation letters considering the facts and law that would
hold good in the court of law if challenged are a major concern.

SOLUTION REQUIRED
Our aim is to develop a solution which will give inputs on specific case details,
material for effective contest tailor made for the case to be made available.
Accurate to the point drafts of various letters like repudiation letters, drafts to be
sent to advocates & investigators etc which would be available by applying the
legal principles basing on the facts of the case.

SUCCESS CRITERIA

• Context based case details are made available.

• Pointed drafts generated with specific inputs on requirements should be
readily available.
 HackRx 4.0

PROBLEM STATEMENT XI

VISUAL TESTING AT SCALE

Our design team has created a visual design language that is followed by the app
and website. This is laid down in the Figma designs that describe how the screens
would look like on app, web (mobile) and web (Desktop).
The website and app screens are developed against this library of components as
documented in Figma. There are predefined rules (font size, padding, number of
text lines, etc.). For every component that define how the component will look on
desktop and mobile viewports.

OBJECTIVE
The objective of this project will be to develop a validation model that will be
trained using Figma, approved screenshots, and component designs. The model
will then be used to rate new screens and pages developed for
compliance/adherence to the Figma design. A human will then review the
deviations and decide if this warrants a bug or not.
This model should be accessible through an API, for this to be integrated with the
testing pipeline. This will enable the QA team to perform visual testing of new
pages quickly, and with minimal manual intervention.
There is tool used in industry – Percy, which does something similar. But the tool
uses screenshots of websites as reference and is very sensitive. Even simple text
changes will be shown as a deviation, whereas the website itself may be compliant.

TYPICAL TECH STACK

Unleash Python, TensorFlow or similar ML modelling platform, Computer Vision
 HackRx 4.0

PROBLEM STATEMENT XII

PROBLEM STATEMENT
Graph data provides a powerful way to represent and identify linkages. How can
the representation (graph data model) with graph machine learning assist in taking
data science applications to enhanced levels is a critical capability. The problem
statement explores how graph representation coupled with graph query (insights)
coupled with graph machine learning (like GNNs. Etc.) can provide next generation
viable business capability. The solution and approach also need to address
optimal/viable architectures to work on optimal memory and overall commodity-
based hardware (servers). This problem will explore an end to end nature on
building scalable graph-based capability with a technical frame using big data, data
pre and post processing, data models for Graph/ML/analytics consumption.
The requirement is to explore these aspects, we use the SNAP product co-
purchasing networks (details provided). The key task is to (first) identify a good
problem to solve that will address some potential business use case (you need to
identify this). This solution (second) needs to include a graph-based approach with
the steps identified in the submission/evaluation section. The final (third)solution
must showcase an integration to a mobile app/ web front end (with API
integration). The data is based on the data based on a large ecommerce platform
and hence some use cases linked to that can also be considered.

TECHNICAL STACK (MANDATORY TO FOLLOW)

Dataset & Technical Stack
Data Set Details Product co- Google collab (Link)
Link Code IDE(s) & Notebook
purchasing networks - SNAP) Visual Studio Code (Link)
Core: Python
Data Set: Meta Data Link Languages & framework Processing: Pyspark | Spark 3.4.0+2.4+
Core Graph Database: (Mandatory to
choose one of the following)
Data Set: To use Link Graph Components JanusGraph
Option 1:
BlazeGraph
Option 2:
Graph Query / Models /API gremlinpython
Visualisation Tools Link
Interface Output:
OUTCOMES EXPECTED
• Post understanding the dataset (both transaction and meta data) identify a
problem that you would want to solve. Clearly articulate the problem statement.
• Data processing (including Loading and transformations) of the data set
(combination of pyspark & python-based code) or other high-performance libraries
(in case being used, need to mention with details)
• The data processing/analysis needs to use both transaction and meta-
data(semi/(un)structured)
• Creation of intermediate and base analytical data models (using both transaction
and meta data)
• Create the graph-based model to help solve your problem statement basis the
standard libraries mentioned; in case you have a better library, please call-out
purpose of using them including providing specifics on them.
• In case the models require post process for computation; please factor the needful
details on how they have been used.
• The final (third)solution must showcase an integration to a mobile app/ web front
end (with API integration).

SUBMISSION & EVALUATION BROAD METRICS (USING

TECHNICAL STACK)
• Problem statement definition (well-defined)
• Pre-processing and data cleaning (including standardising)
• Building reference/base data sets that facilitate modelling
• Exploratory Data Analysis (linked to problem statement)
• Data Processing pipelines
• Model Build Approach & Selection (includes testing) (using Graph based approach)
• Model Execution |Results & relevance as per problem statement
• Integration Approach | Mobile Application or Web Application
• Model Management and Performance Tracking Approach
• Guidance to Code level
a) Inline | Code | Documentation | Markdowns
b) High Quality (modular, coding guidelines, etc.)
c) Optimisations and efficiency codified basis your assignment
i. Algorithm/Time complexity, scalability & efficiency)
ii. Performance (time, memory metrics, etc.)
iii. Integrations

SUBMISSION SPECIFICS

• Must be a notebook / project code (runnable live) as per platforms indicated above.
• Notebook must be named:
HealthRx40_Hackathon_23_<NameOfTeam_Person>
• Additional Document (if required):
HealthRx40_Hackathon_23_<NameOfTeam_Person>.doc
• Additional aspects can also be covered in the notebook as markdown or additional
document can also be considered.
• Code must be high quality and shared for evaluation