Thadomal Shahani Engineering College
Problem statement and
instructions
Module
Google Developer Groups
Thadomal Shahani Engineering College
Problem Statement
and
Instruction Module
Google Developer Groups
Thadomal Shahani Engineering College
1. It is compulsory for all teams to include
atleast one Google technology. Hosting your
project on firebase also counts for the
same.
2. You have to make a ppt in order to present
it in front of the judges.
3. Judging will be held for students who
submit their projects and ppt by 7.30 pm on
the sent site.
4. Decision of the judges remain final.
Thadomal Shahani Engineering College
Project AETHER
Overview
Project AETHER is a deliberative multi-agent AI system that analyzes
complex reports by breaking them into key factors and evaluating each
through structured debate between opposing agents. For every factor, one
agent presents supportive arguments while another critically challenges
them, directly responding to each other’s claims. A synthesizer agent then
combines these debates into a single transparent, actionable report
explaining what worked, what failed, why it happened, and how to improve,
ensuring conclusions are based on multi-agent reasoning rather than a
single model across domains such as sales, statistics, organizational
analysis, and policy evaluation.
Required Features
Use multiple AI agents with clearly defined responsibilities, including
factor extraction, supportive reasoning, Opposing REASONING, and final
Decision Making.
Demonstrate explicit agent-to-agent interaction, where arguments are
challenged, defended, and refined through direct responses.
Implement a structured coordination flow that governs how agents
communicate, exchange context, and progress through reasoning stages.
Generate a single unified final report that meaningfully integrates all
agent perspectives instead of presenting isolated model outputs.
Ensure transparency and interpretability by exposing reasoning logs,
debate traces, or execution-flow visualizations.
Bonus Features (Optional)
Additional credit will be awarded for:
Deploy each agent independently (locally or in the cloud) with real-time
communication via defined coordination mechanisms.
Support multimodal inputs, including text, structured data, charts, or
other formats.
Thadomal Shahani Engineering College
Project MIRAGE
Overview
Project MIRAGE focuses on advancing human–AI interaction by moving
beyond static, text-only chatbots to lifelike, expressive AI avatars. The
goal is to create a digital human clone capable of listening, reasoning,
speaking, and visually reacting in a coherent and emotionally engaging
manner Rather than prioritizing perfect photorealism, MIRAGE emphasizes
on synchronized multimodal interaction—where speech, facial animation,
and expressive feedback align naturally with the AI’s reasoning. These
avatars can act as teachers, assistants, customer service agents, or
companions, helping reduce the emotional and communicative gap between
humans and machines through real-time, human-like interaction.
Required Features
Your system must:
Accept user input via text, speech, or both.
Generate intelligent responses using natural-language reasoning
(LLM, RAG, or rule-based approaches are acceptable).
Convert AI-generated responses into audio using text-to-speech or
synthesized voice.
Display a visible avatar or human representation that reacts
through facial expressions, motion, or gestures.
Maintain real-time responsiveness to ensure smooth conversational
flow without noticeable delays or interruptions.
Bonus Features (Optional)
Additional credit will be awarded for:
Emotion mapping: Detecting user sentiment or contextual cues and
adjusting the avatar’s tone, facial expressions, or gestures
accordingly.
Personalized memory: Retaining conversational context, preferences,
or user-specific information across sessions to enable more
personalized interactions.
Thadomal Shahani Engineering College
Project SENTINEL
Overview
Millions of people with visual, auditory, or mobility impairments face
challenges such as navigating environments, interpreting visual cues, or
recognizing critical sounds. Project SENTINEL addresses this by using AI to
transform visual data into real-time, actionable assistance.
Participants must build an AI-powered assistive vision PWA (Progressive
Web app) / NATIVE APP that analyzes live or recorded visual input with
computer vision and machine learning, identifies objects, text, gestures,
obstacles, or environmental cues, and delivers accessible feedback
through audio, visual, or haptic channels to help users better
understand and interact with their surroundings.
Required Features
Your system must:
Capture and process live or recorded video input from a smartphone
or camera module.
Use computer vision or machine learning models to detect, identify, or
interpret relevant objects, actions, or environmental cues.
Provide real-time feedback through at least one output channel,
such as audio, vibration, visual display, or text-to-speech.
Target a specific disability context, including visual impairment,
hearing impairment, or mobility assistance.
Demonstrate reliable detection and response accuracy in varied or
real-world-like conditions.
Bonus Features (Optional)
Additional credit will be awarded for:
Integration with wearable devices — smart glasses, wristbands, or
AR/VR headsets.
Voice-enabled interaction — allowing users to ask questions about
their surroundings.
Accessibility across multiple disabilities — combining vision, sound
recognition, and speech output.
Thadomal Shahani Engineering College
Project SKILLSPHERE
Overview
In a rapidly evolving job market, students and early professionals often
struggle with career planning, skill identification, and job preparation.
Project SKILLSPHERE addresses this challenge by providing an AI-powered
career guidance platform that acts as a personal mentor. By analyzing
a user’s skills, education, interests, and goals, the system helps explore
suitable career paths, build job-ready resumes, assess readiness, and
receive personalized learning recommendations, making career
development more structured, interactive, and accessible.
Required Features
AI Career Path Recommender: Suggests suitable career paths and
trending roles by analyzing user skills, education, and interests, with
guidance on fast-growing industries.
AI Resume Builder & Readiness Evaluator: Generates and reviews
resumes, improving structure, formatting, and keywords for ATS
compatibility and job readiness.
Personalized AI Learning Guide: Identifies skill gaps and recommends
relevant courses or certifications, updating suggestions as the user
progresses.
Interactive Interview Preparation Module: Provides a live interview
simulator with video, screen sharing, coding support, and AI-driven
feedback, including mock interviews with peers or mentors.
Bonus Features (Optional)
Additional credit for implementing:
Career Persona Generator: Creates a summarized professional
profile from user data.
Portfolio Builder: Automatically compiles projects into a shareable
portfolio.
Peer Learning Circles: Enables collaboration through mock interviews
or study groups.
Job Trend Tracker: Shows real-time insights on role demand and
salary trends.
Thadomal Shahani Engineering College
Project INTEGRITY
Overview
Citizens often face inconsistencies in public infrastructure, such as roads
failing soon after repairs or facilities existing only in official records. While
public spending data is available, it is buried in complex documents, making
verification difficult. Project INTEGRITY addresses this by providing a web and
mobile platform that uses Agentic AI and geospatial visualization to help
citizens report issues, verify public spending, and view the overall “health” of
a municipality through an interactive, transparent map—bridging the gap
between official data and on-ground reality.
Required Features
Automated Document Analysis: Processes government PDFs or images using AI
(LLMs) to extract key metadata such as project name, budget, and contractor.
Visual Evidence Verification: Allows users to capture or upload infrastructure
images with automatic GPS and timestamp logging, while computer vision classifies
issues (e.g., potholes or incomplete construction).
Agentic Audit Workflow: An autonomous AI process with three stages:
1. Retrieval: Fetches relevant project data from a structured
dataset (mock database allowed).
2. Audit: Compares user evidence with official records to detect
discrepancies.
3. Verdict: Generates a structured risk level with supporting reasoning.
Geospatial Corruption Visualization: Uses an interactive map (e.g., Leaflet or
Mapbox) with clustering or heatmaps to highlight compliant and high-
discrepancy areas.
Categorical Filtering: Filters map data by infrastructure type such as roads,
sanitation, or public works.
Bonus Features (Optional)
Additional credit will be awarded for:
Privacy-First Reporting: Anonymous submissions to protect user identity.
Automated Correspondence: AI-generated complaint letters or RTI applications
from verified reports.
Predictive Analytics: Forecasts potential infrastructure issues using report
and spatial trends.
Thadomal Shahani Engineering College
Project BRANDPULSE
Overview
Launching an effective brand campaign requires strategic planning, creative
design, compelling copy, and precise media execution, often a bottleneck for
startups and marketing teams. Project BRANDPULSE addresses this challenge by
building an autonomous AI platform that acts as a complete Brand Campaign
Strategist. The system ingests a high-level user brief and autonomously
reasons through strategy, research, and creative generation, orchestrating
multiple AI tools to produce a cohesive, deployment-ready campaign package
for review and launch.
Required Features
Abstract Reasoning & Strategy Formation: Interprets and breaks down a user’s high-
level request to analyze the target audience, context, and product, then
formulates a core campaign strategy or theme.
Orchestration of Specialized Generative Tools: Based on the defined strategy, the
system coordinates multiple generative tasks:
1. Visual Identity: Uses image generation models to create mood
boards and key visual assets.
2. Copywriting: Generates social media captions, ad copy, and blog content
aligned with the campaign theme.
3. Market Research: Identifies relevant influencers or Publications and
drafts personalized outreach messages.
4. Media Planning: Recommends posting schedules and channel strategies
for optimal reach.
Interactive Campaign Canvas: Presents the complete campaign within an interactive
web-based workspace where users can review, edit, regenerate assets, and export
the final campaign package.
Bonus Features (Optional)
Additional credit will be awarded for implementing:
Competitor Insight Dashboard: Analyzes competitor campaigns for strategic
differentiation.
Campaign Performance Predictor: Predicts reach, engagement, and overall
effectiveness.
Team Collaboration Mode: Supports real-time review and iteration by multiple
stakeholders.
Thadomal Shahani Engineering College
Project VEIL
Overview
Many crimes go unreported due to fear of exposure, retaliation, or loss
of privacy. Traditional centralized reporting systems often suffer from
critical weaknesses, including censorship risks, data tampering, identity
leaks, and single points of failure. Project VEIL addresses these
challenges by proposing a decentralized, blockchain-powered crime
reporting platform that enables citizens to submit reports anonymously
while preserving data integrity and authenticity. The system is designed
to balance strong anonymity with trust and accountability, encouraging
genuine reporting while actively discouraging false or malicious
submissions through cryptographic and incentive-based mechanisms.
Required Features
Your system must:
Enable encrypted and anonymous crime report submissions without
revealing user identity.
Store reports immutably using blockchain or distributed storage
technologies.
Use smart contracts to verify reports and reward valid submissions.
Implement penalty mechanisms to deter spam, false, or malicious
reports.
Support secure, end-to-end encrypted communication between
reporters and authorized authorities.
Ensure sensitive information is protected using robust cryptographic
techniques.
Bonus Features (Optional)
Additional credit will be awarded for implementing:
Reputation-Based Validation: Systems that evaluate report
credibility over time based on contributor behavior.
AI-Driven Content Filtering: Automated detection of spam, abuse, or
low-quality submissions using AI models.
Thadomal Shahani Engineering College
Project FORTRESS
Overview
Modern financial applications face growing risks from cyberattacks,
data breaches, and unauthorized access. Traditional centralized
architectures expose sensitive financial data to single points of
failure, making large-scale compromise more likely. Project FORTRESS
addresses these challenges by leveraging blockchain technology to
build a secure FinTech platform focused on data confidentiality,
transaction integrity, and robust access control. By combining
decentralization, encryption, and smart contract automation, the
platform aims to create a transparent, tamper-resistant, and
resilient financial security system capable of operating in high-risk
environments.
Required Features
Your system must:
Implement end-to-end encryption for user data and financial
transactions.
Use smart contracts to automate and secure financial
operations.
Support decentralized authentication and identity verification
mechanisms.
Enforce role-based or attribute-based access control for
sensitive actions.
Detect anomalies or fraudulent behavior in real time using
monitoring or analytics mechanisms.
Bonus Features (Optional)
Additional credit will be awarded for implementing:
Immutable Audit Trails: Tamper-proof transaction logs that
support regulatory compliance and post-incident verification.
Thadomal Shahani Engineering College
Project BALLOT
Overview
Elections worldwide continue to face challenges related to trust,
vote manipulation, lack of transparency, and low voter confidence.
Centralized electronic voting systems are particularly vulnerable
to tampering, hacking, and opaque result handling, raising concerns
about election integrity. Project BALLOT addresses these issues by
proposing a decentralized, blockchain-based voting system that
ensures voter privacy, prevents fraud, and enables transparent,
verifiable election outcomes. By leveraging cryptographic techniques,
smart contracts, and distributed ledger technology, the system aims
to build trust in digital voting while safeguarding voter anonymity.
Required Features
Your system must:
Support secure voter authentication using mechanisms such as
national ID verification or OTP-based validation.
Record each vote as a unique, tamper-proof on-chain transaction.
Ensure anonymous yet verifiable storage of votes.
Use smart contracts to automate vote counting and publish
results transparently.
Perform AI-based face detection and image comparison using
pretrained computer vision models (e.g., YOLO), demonstrated on
sample or simulated identity datasets.
Bonus Features (Optional)
Additional credit will be awarded for implementing:
Zero-Knowledge Proofs: Cryptographic mechanisms to further
enhance voter anonymity.
Offline Vote Caching: Support for voting in low-connectivity or
offline environments with later synchronization.