Browse My Recent

Projects

MediAssist+

React Native Node.js PostgreSQL Local AI

An AI-powered mobile app for doctors that records consultations, converts speech to text, and generates structured medical summaries.

MediAssist+ - AI-Powered Consultation System

MediAssist+ is an advanced mobile application for healthcare professionals that records consultations, converts speech to text, and generates structured medical summaries using local AI models.

System Architecture

Mobile Client: Developed in React Native, managing audio buffers and doctor navigation routes.
Backend API: Node.js and Express gateway checking doctor tokens.
Database: Secure PostgreSQL schemas storing history under HIPAA alignment.
AI Engine: Python FastAPI microservice hosting Whisper for transcription and local Llama 3 models for clinical note summaries.

Key Features

High-fidelity real-time consultation voice recording and conversion to text.
AI clinical note generator outputting SOAP summaries (Subjective, Objective, Assessment, Plan).
Secure doctor dashboard log with search, filters, and profile controls.
Automated patient files deletion and secure session logs.

Technical Challenges Resolved

Problem: Deciphering complex medical terms and jargon.
Solution: Fine-tuned model sampling parameters and integrated a post-transcription custom dictionary matching phase.

Future Scope

Integration with Electronic Health Records (EHR) databases via FHIR APIs.
Multi-lingual transcription translation.

AI Fruit Freshness Detection

TensorFlow Keras Flask Python

A deep learning application to classify fruit freshness and detect spoilage in real-time using image data.

AI-Based Fruit Freshness Detection System

An intelligent computer vision system classifying fruit freshness and detecting early-stage spoilage using deep learning CNNs and serving predictions via a Flask API.

System Architecture

Web Client: Drag-and-drop file upload views built with HTML, CSS, and JS.
Inference API: Flask web backend loading model weights.
Deep Learning Core: TensorFlow/Keras Convolutional Neural Network trained on optimized fruit classification datasets.

Key Features

Real-time image classification for multiple fruit types.
Visual metrics display indicating freshness confidence levels.
Spoilage timeline estimates based on current spoilage stages.
Batch processing inputs processing multiple items at once.

Technical Challenges Resolved

Problem: Model precision drops due to lighting changes in different environments.
Solution: Augmenting the training dataset with random shadows, contrast shifts, and brightness scaling.

Future Scope

Deploying the model as a lightweight TensorFlow Lite container inside mobile applications.
Integrating automated warnings for inventory control frameworks.

Automated OMR Evaluation

OpenCV Flask Computer Vision Python

An automated computer vision grading application that processes and scores scanned bubble sheets with extreme accuracy.

Automated OMR Evaluation & Scoring System

This system automates the evaluation of OMR sheets using advanced computer vision and web APIs. Leveraging OpenCV for image transformation and Flask for backend services, users can upload scanned sheets and instantly compute scores against a CSV answer key.

System Architecture

UI Dashboard: Interface for student roster management and CSV grading key uploads.
Backend API: Flask file receiver handling multi-part scanner files.
Computer Vision Core: OpenCV transformation pipeline executing perspective correction and bubble analysis.

Key Features

Automatic contour alignment and perspective transform for warped or rotated scans.
Intelligent bubble grid detection based on threshold density analysis.
Rapid scoring engines (processing sheets in milliseconds).
Excel/CSV roster exports for graded sheets.

Technical Challenges Resolved

Problem: Tilt variations and camera distances led to misalignment.
Solution: Implemented Hough Line Transforms to detect paper boundaries and warps, ensuring accurate cell alignment.

Future Scope

Adding Optical Character Recognition (OCR) for hand-written student roll numbers.
Multi-page exam booklet packet scanning.

Crop Disease Detection

TensorFlow Keras OpenCV Flask

A deep learning classification system that detects plant leaf diseases from images using CNNs and provides remedies.

Crop Disease Detection Using Deep Learning

A high-fidelity machine learning system that uses a CNN to detect plant diseases directly from leaf images. The model flags potential crop diseases and suggests effective remedies via a real-time prediction API.

System Architecture

Client Interface: Lightweight dashboard showing scan reports.
API Server: Flask endpoint forwarding image files to model loader.
ML Classifier: MobileNetV2 architecture fine-tuned on PlantVillage dataset.

Key Features

Instant classification of crop diseases (leaf blights, rusts).
Organic and chemical remedies database logs.
Lightweight model footprint supporting rapid processing.
Detailed history logs of scanned plant diseases.

Technical Challenges Resolved

Problem: Model parameters were too large for fast loading.
Solution: Quantized model weights to float16 using TensorFlow Lite converter, decreasing file footprint by 75%.

Future Scope

Offline local predictions on native Android application.
Geolocated disease outbreak warning alerts.

Vaidra

FastAPI Flutter Docker PostgreSQL

A comprehensive engineering project featuring FastAPI backend services, a Flutter mobile app, and containerized Docker setup.

Vaidra - Multi-Component Architecture

Vaidra is a comprehensive engineering project featuring a multi-component architecture, including a Python-based backend (FastAPI), a Flutter mobile application, and Docker-based deployment.

System Architecture

Client App: Cross-platform Flutter Mobile Application.
Backend API: FastAPI REST services managing models.
Database Layer: PostgreSQL storage with Alembic schema management.
Deployment: Multi-stage containerized setup managed via Docker Compose.

Key Features

Secure JWT user authentication and route verification.
Optimized SQL queries using SQLAlchemy ORM.
Automated schema updates and migrations using Alembic.
Consistent development environments using containerization.

Technical Challenges Resolved

Problem: Running Alembic migrations before database initialization threw errors.
Solution: Created entrypoint scripts checking connection queries before launching Alembic migrations.

Future Scope

Kubernetes configuration scripts for auto-scaling.
Real-time state update servers using WebSockets.

ResumeRelevance

Python Streamlit NLP LLM

An AI-driven application designed to analyze and score resume relevance against technical job descriptions.

ResumeRelevance - AI Resume Scoring

An AI-driven application designed to analyze and score resume relevance against technical job descriptions.

System Architecture

Client Views: Streamlit interactive uploads dashboard.
Processing Core: Python PyPDF2 extraction parser.
AI Scoring Model: NLP SentenceTransformers embeddings using Cosine Similarity calculations.

Key Features

Direct parsing pipelines for PDF/DOCX files.
Real-time relevancy score calculation.
Dynamic listing of missing keywords and skills.
Formatting tips dashboard for candidates.

Technical Challenges Resolved

Problem: Parsing text from multi-column resume PDFs resulted in scrambled layouts.
Solution: Implemented linear bounding-box section parsers to extract columns in correct sequential order.

Future Scope

Integrating LLM API calls for automated rewrite advice.
Auto-fetching job descriptions via LinkedIn scraping.

Task Manager with Progress Tracker

HTML5 CSS3 JavaScript LocalStorage

An intuitive web productivity application featuring interactive progress trackers, category management, and local persistence.

Task Manager & Productivity Suite

An intuitive web productivity application featuring interactive progress trackers, category management, and local persistence.

System Architecture

Interface: Styled HTML5 and CSS3 Layout.
Controller: Vanilla Javascript DOM event mapping logic.
Data Layer: Browser LocalStorage persistence hooks.

Key Features

Interactive checklists with visual completion counters.
Stepless progress rings updating on state updates.
Separate categories for Work, Study, and Personal items.
Persistent data preventing loss on browser resets.

Technical Challenges Resolved

Problem: Real-time progress indicators failed to update when items were deleted.
Solution: Unified state managers triggering counter updates on all list adjustments.

Future Scope

Calendar views and time tracking cards.
Cloud synchronization integrations.

Recipe Book

HTML5 CSS3 JavaScript

A gorgeous web application showcasing recipes with high-fidelity visual design, interactive categories, and searchable catalog.

Recipe Book & Search Portal

A gorgeous web application showcasing recipes with high-fidelity visual design, interactive categories, and searchable catalog.

System Architecture

Frontend: Responsive layout built with CSS grids.
Controller: Vanilla JS filtering search layer.

Key Features

Instant search filtering indexing.
Popup recipe views displaying details and prep times.
Interactive category tag switches.
Clean responsiveness on mobile screen resolutions.

Technical Challenges Resolved

Problem: Typing queries quickly caused sluggish response rendering.
Solution: Debounced query listener inputs to optimize browser reflow times.

Future Scope

User ratings system.
Shopping lists ingredient compilation.

Chandrashekhar M Loni

About Me

Experience

Education

Journey

10th Grade

Jawahar Navodaya Vidyalaya, Vijayapur

12th Grade

Jawahar Navodaya Vidyalaya, Vijayapur

B.E. in Artificial Intelligence & Machine Learning

SDMCET Dharwad

Front-End Web Dev Intern

Edunet Foundation (AICTE & IBM SkillsBuild)

App Development Intern

BlackSpire Pvt. Ltd.

Skills

Technical Skills

HTML

CSS

JAVASCRIPT

C / C++

Python

Java

SQL

React

Android

Projects

MediAssist+

MediAssist+ - AI-Powered Consultation System

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

AI Fruit Freshness Detection

AI-Based Fruit Freshness Detection System

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

Automated OMR Evaluation

Automated OMR Evaluation & Scoring System

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

Crop Disease Detection

Crop Disease Detection Using Deep Learning

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

Vaidra

Vaidra - Multi-Component Architecture

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

ResumeRelevance

ResumeRelevance - AI Resume Scoring

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

Task Manager with Progress Tracker

Task Manager & Productivity Suite

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

Recipe Book

Recipe Book & Search Portal

System Architecture

Key Features

Technical Challenges Resolved

Future Scope

Contact Me

Email

LinkedIn

Instagram