Student Transformation Stories

From confusion
to confidence.

Every story here is real. Names, colleges, projects — all real. These are students from Anurag College of Engineering and Vagdevi College of Engineering who came to Visthar with borrowed projects and left having built something they truly understand.

R

Roopa

Anurag College of Engineering · Batch 2026

AI / Machine Learning
CS_001

Project

HealthAI Portal: Cardiovascular Risk Prediction Using Random Forest & LLM-Generated Recommendations

Before Visthar
  • Had a basic health monitoring app with no ML model — just UI with dummy data
  • Couldn't explain how a Random Forest model actually makes predictions
  • No understanding of feature selection or model evaluation metrics
  • No idea how to integrate an LLM into a healthcare workflow
After Visthar
  • Built a full Random Forest pipeline from scratch — data preprocessing to prediction
  • Integrated an LLM for generating personalised health recommendations
  • Implemented proper evaluation metrics: AUC-ROC, Precision, Recall, F1-score
  • Defended every architectural decision — feature importance, model selection, LLM prompt design
PythonRandom ForestLLMNLPFlaskScikit-learn

“Visthar made me understand every algorithm I used — not just the code, but why Random Forest was the right choice. I could explain the math behind the risk scoring during my Viva.”

— Roopa, Anurag College of Engineering

Outcome

Project Accepted — Research Paper Written for IEEE

V

Vinay

Anurag College of Engineering · Batch 2026

Cloud / Security / AI
CS_002

Project

ArchSecure: Cloud Security Framework Selection Using Weighted Deterministic Scoring & LLM-Based Advisory

Before Visthar
  • Had a generic 'cloud security dashboard' — no real evaluation logic
  • Couldn't explain what makes one security framework better than another
  • No understanding of weighted scoring models or how to build one
  • LLM integration felt impossible — didn't know where to start
After Visthar
  • Designed a weighted deterministic scoring model for framework evaluation
  • Built an LLM-based advisory layer that explains why a framework was recommended
  • Implemented proper scoring logic with adjustable weights and justification logs
  • Defended the entire evaluation pipeline during project review — guide was thoroughly impressed
PythonLLMCloud APIsFlaskSecurity StandardsREST APIs

“The LLM-based advisory part was something I never thought I could pull off. Siddi showed me how to design the evaluation pipeline so it's defensible from every angle.”

— Vinay, Anurag College of Engineering

Outcome

Project Selected for College Innovation Expo

A

Abhishikth

Anurag College of Engineering · Batch 2026

Blockchain / Web3
CS_003

Project

Algorand-Based Blockchain Solution for Transparent Record Keeping

Before Visthar
  • Had a basic blockchain demo copied from online — couldn't explain how it worked
  • No understanding of how Algorand differs from Ethereum or Bitcoin
  • Smart contracts were a black box — copied from tutorials without comprehension
  • No idea about consensus mechanisms or transaction validation
After Visthar
  • Understanding of Algorand's Pure Proof of Stake consensus from first principles
  • Designed and deployed custom smart contracts on Algorand Testnet
  • Built a complete record-keeping dApp with wallet integration and transaction history
  • Explained every component of the blockchain architecture — from nodes to transaction validation
AlgorandPythonSmart ContractsBlockchainFlask

“Blockchain projects are complex by nature — but Visthar made me understand every smart contract I wrote. When my guide asked about consensus mechanisms, I answered confidently.”

— Abhishikth, Anurag College of Engineering

Outcome

Project Accepted — Faculty Called it Technically Rigorous

AN

Anirudh

Anurag College of Engineering · Batch 2023

AI / Interpretable ML
CS_004

Project

Predicting & Understanding College Student Mental Health with Interpretable Machine Learning

Before Visthar
  • Had a standard classification model with no interpretability layer
  • Couldn't explain why the model made specific predictions for individual students
  • No understanding of XAI (Explainable AI) concepts or tools like LIME or SHAP
  • Project lacked any academic depth for research paper writing
After Visthar
  • Designed an interpretable ML pipeline using LIME and SHAP for local explanations
  • Built model-agnostic explanation system that generates per-prediction reports
  • Wrote the architecture section of a research paper on interpretable ML in education
  • Defended the XAI approach in project review — very few students in the department were doing this
PythonXAIScikit-learnFlaskLIMESHAP

“Interpretable machine learning was the hardest part. Visthar helped me design explanations for every prediction — not just the model, but why it made that specific decision for a specific student.”

— Anirudh, Anurag College of Engineering

Outcome

Research Paper Written for IEEE

I

Irfan

Anurag College of Engineering · Batch 2026

IoT / AI / Healthcare
CS_005

Project

Edge-Enabled IoMT Heart Disease Detection Using Federated Learning

Before Visthar
  • Had a basic heart rate monitor using ESP32 — no ML component
  • No understanding of federated learning or how it differs from centralised training
  • Struggling to figure out how to run a model on edge hardware
  • No idea how to connect firmware, database, and ML inference into one pipeline
After Visthar
  • Designed a federated learning pipeline where ESP32 nodes train locally and share gradients
  • Implemented model compression and quantisation to run on edge devices
  • Built the complete pipeline: sensor data collection → model inference → results display
  • Wrote a research paper documenting the federated learning architecture for IoMT
ESP32Federated LearningPythonIoMTMQTTTensorFlow

“Federated learning on edge devices was complex — Siddi broke it down step by step. From sensor data collection to model inference, I built everything myself.”

— Irfan, Anurag College of Engineering

Outcome

Research Paper Written for IEEE — Project Accepted

HK

Harshan Kumar

Anurag College of Engineering · Batch 2026

Computer Vision / Deep Learning
CS_006

Project

An Attention-Based Feature Processing Method for Cross-Domain Hyperspectral Image Classification

Before Visthar
  • Had a basic CNN image classifier — no understanding of attention mechanisms
  • Couldn't explain why attention helps in hyperspectral image analysis
  • No mathematical grounding for how attention weights are computed
  • Research paper seemed impossible — didn't know how to write methodology
After Visthar
  • Implemented self-attention mechanism from scratch — understood Q, K, V matrices deeply
  • Designed triple-attention fusion module: channel, spatial, and pixel-level attention
  • Wrote a complete research paper documenting the methodology for IEEE
  • Explained transformer architecture and attention math during project defense — guide was genuinely impressed
PythonPyTorchCNNAttention MechanismsNumPy

“Hyperspectral image classification requires strong mathematical grounding. Visthar helped me understand attention from the ground up — not copied a tutorial. I explained the transformer architecture in my Viva and my guide was impressed.”

— Harshan Kumar, Anurag College of Engineering

Outcome

Research Paper Written for IEEE

G

Goutham

Anurag College of Engineering · Batch 2023

AI / Healthcare / Deep Learning
CS_007

Project

An Improved Infectious Disease Risk Prediction Model Based on Triple-Attention Fusion

Before Visthar
  • Had a basic disease prediction model with poor accuracy
  • Didn't understand how to fuse different types of attention mechanisms
  • No experience writing research papers — no idea how to structure one
  • Struggling to connect the ML model to a frontend application
After Visthar
  • Implemented triple-attention fusion architecture with channel, spatial and temporal attention
  • Significantly improved model accuracy through proper attention design and hyperparameter tuning
  • Built a complete web application with model inference API and result visualisation
  • Wrote the complete research paper with methodology, experiments, and results — submitted for IEEE
PythonTensorFlowTriple-AttentionFlaskPandasNumPy

“Triple-attention fusion was a concept I read in a research paper — Visthar helped me implement it from scratch and write about it academically. The paper that came out of this project was a huge confidence boost.”

— Goutham, Anurag College of Engineering

Outcome

Research Paper Written for IEEE

R2

Rahul

Vagdevi College of Engineering · Batch 2026

IoT / Full-Stack / Startup
CS_008

Project

Vellora — Government Bus Tracking & Logging System (Startup)

Before Visthar
  • Had a basic idea for a bus tracking system — no prototype, no technical direction
  • No idea how to build hardware that tracks a vehicle in real time
  • Didn't know how to design the database for a real-time fleet tracking system
  • No path from college project to actual product — felt like the idea would stay an idea forever
After Visthar
  • Built the first working prototype of Vellora — ESP32-based bus tracker hardware
  • Designed and implemented the complete database architecture for fleet tracking
  • Developed the application that connects the tracker hardware to real-time data display
  • First prototype complete — now in active conversations with government officials
ESP32MQTTDatabaseApplication DevelopmentReal-Time Tracking

“Visthar helped me build the first real prototype of Vellora. I built the hardware tracker, connected it to a database, and delivered a working application. Now we're talking to government officials. Visthar helped me build something real.”

— Rahul, Vagdevi College of Engineering

Outcome

First Prototype Complete — Startup Growing (Vellora)

Want to be our next case study?

Your story could be the one that shows the next Dean or HOD what Visthar really does. Let's build it together.

Start Your Journey