HS

Harshit SoniAI-ML Enthusiast & Fullstack Developer

A passionate AI-ML enthusiast and full-stack developer with a strong foundation in computer science and a drive to create innovative solutions.

PythonPyTorchTensorFlowJavaScriptNext.jsReactNode.jsMongoDB

Loading about section...

Featured Projects

A curated selection of my work, showcasing my skills in creating modern, user-centric applications.

Q & A Bot for College Admission
Python
NLTK
Numpy
Q & A Bot for College Admission
The Basic Q&A Bot for College Admission is a Python-based chatbot that answers queries about admission procedures, requirements, and deadlines through a Tkinter-powered interface. It supports contextual multi-turn conversations, error handling, and is built with NLTK and a trained model for intelligent responses. Future upgrades include backend integration and advanced NLP features.
Python
NLTK
Numpy
TensorFlow
Keras
Pickel
Tkinter
Python
NLTK
Numpy
TensorFlow
Keras
Pickel
Tkinter
Crisis Management Platform
HTML
CSS
JavaScript
Crisis Management Platform
Crisis Management Platform is a hackathon-built web app featuring an AI model for fire detection via IP camera and a weather report API to boost situational awareness. Led by a team of six, it achieved over 95% fire detection accuracy and plans to expand with alert systems and rainfall prediction.
HTML
CSS
JavaScript
Bootstrap
Python
Flask
OpenCV
Tensorflow
Keras
HTML
CSS
JavaScript
Bootstrap
Python
Flask
OpenCV
Tensorflow
Keras
Self Driving Car (JS)
HTML
CSS
JavaScript
Self Driving Car (JS)
Self Driving Car (JS) is a browser-based simulator that uses JavaScript to model autonomous vehicle behavior. Users can train the car by saving sensor parameters, tweak simulation settings like sensor count and car numbers, and visualize learning iterations live in the browser.
HTML
CSS
JavaScript
Reinforcement Learning
HTML
CSS
JavaScript
Reinforcement Learning
ChatWithWebsite
Next.js
TailwindCSS
Upstash Redis
ChatWithWebsite
ChatWithWebsite is a web app that lets users interact with any website by entering its URL into a message box, enabling quick redirection and information retrieval. It features a clean UI, seamless navigation, and responsive design for all devices.
Next.js
TailwindCSS
Upstash Redis
Rag Chat
TypeScript
Next.js
TailwindCSS
Upstash Redis
Rag Chat
TypeScript
VisionMATE: Assistive Technology for Visually Impaired
React.js
TensorFlow.js
Web Speech API
VisionMATE: Assistive Technology for Visually Impaired
VisionMATE is a full-featured web app designed to support visually impaired students through real-time object detection, text-to-speech, speech-to-text, and voice-controlled task management. Built with React.js, TensorFlow.js, and Node.js, it offers a responsive, accessible interface powered by motion and speech technologies.
React.js
TensorFlow.js
Web Speech API
TailwindCSS
Node.js
Express
MongoDB
JWT Auth
Python
Yolo
OpenCV
React.js
TensorFlow.js
Web Speech API
TailwindCSS
Node.js
Express
MongoDB
JWT Auth
Python
Yolo
OpenCV
Two-Wheeler Road Surface Classifier
Python
MIT App Inventer
LSTM
Two-Wheeler Road Surface Classifier
An Android app that records accelerometer and gyroscope data, builds a dataset, and trains an LSTM to classify 3‑second samples into Kankar, Bitumen, Concrete, single speed breaker, and multiple speed breakers. It enables on-bike, real-time surface classification and can be extended to rash-driving alerts using the same motion dataset.
Python
MIT App Inventer
LSTM
Pandas
Numpy
TensorFlow
Python
MIT App Inventer
LSTM
Pandas
Numpy
TensorFlow
Personal Portfolio
NextJs
MongoDB
Vercel
Personal Portfolio
This is a completely data-driven website that fetches data from MongoDB and images from Cloudinary and has an admin panel to control all this data.
NextJs
MongoDB
Vercel
Cloudinary
NextJs
MongoDB
Vercel
Cloudinary

All Projects

MS Reward Point Collector
Python
Selenium
Pytrends
MS Reward Point Collector
The MS Reward Point Collector is a Python utility script that automates Microsoft Rewards point collection using Selenium and Edge browser. After manual login, it performs trending searches across multiple accounts simultaneously, then closes all windows—maximizing efficiency and time savings.
Python
Selenium
Pytrends
TQDM
Python
Selenium
Pytrends
TQDM
YT Playlist Arranger
Python
Selenium
Time
YT Playlist Arranger
The YT Playlist Arranger is a Python utility that automates renaming downloaded YouTube playlist videos by adding serial numbers. It uses findTitle.py to extract video titles via Selenium and save them to a CSV, then rename.py
Python
Selenium
Time
CSV
OS
Shutil
Python
Selenium
Time
CSV
OS
Shutil
Q & A Bot for College Admission
Python
NLTK
Numpy
Q & A Bot for College Admission
The Basic Q&A Bot for College Admission is a Python-based chatbot that answers queries about admission procedures, requirements, and deadlines through a Tkinter-powered interface. It supports contextual multi-turn conversations, error handling, and is built with NLTK and a trained model for intelligent responses. Future upgrades include backend integration and advanced NLP features.
Python
NLTK
Numpy
TensorFlow
Keras
Pickel
Tkinter
Python
NLTK
Numpy
TensorFlow
Keras
Pickel
Tkinter
Fire Detection Model
Python
OpenCV
Numpy
Fire Detection Model
Fire Detection Model is a Python-based project showcased at SITE-Hackathon 2024, designed to detect fire using computer vision. It uses a Kaggle-sourced dataset and includes training and testing scripts (train.py, testcam_*.py) for real-time fire detection via camera input. The model achieves high accuracy and is structured for easy deployment.
Python
OpenCV
Numpy
Keras
TensorFlow
Scikit-Learn
Python
OpenCV
Numpy
Keras
TensorFlow
Scikit-Learn
Gender-Detection-Master
Python
TensorFlow
Numpy
Gender-Detection-Master
Gender Detection Master is a Python-based project that uses facial recognition to classify gender from webcam input. It includes a pre-trained model, real-time detection script (detect_gender_webcam.py), and training pipeline, and was showcased at the Best Department Contest.
Python
TensorFlow
Numpy
OpenCV
CVLib
Matplotlib
Scikit-Learn
Glob
Python
TensorFlow
Numpy
OpenCV
CVLib
Matplotlib
Scikit-Learn
Glob
Disease-Prediction-System-using-Machine-Learning
Python
Numpy
Pandas
Disease-Prediction-System-using-Machine-Learning
The Disease Prediction System uses machine learning to analyze health data and predict the likelihood of heart disease or diabetes, based on datasets from UCI and Kaggle. It features data preprocessing, feature selection, model training (Logistic Regression, Random Forest, SVM), and a Streamlit-based UI for user input and predictions. Integration with EHR and SHAP-based interpretability enhance its practical utility.
Python
Numpy
Pandas
Scikit-Learn
Streamlit
Pickle
Python
Numpy
Pandas
Scikit-Learn
Streamlit
Pickle
Crisis Management Platform
HTML
CSS
JavaScript
Crisis Management Platform
Crisis Management Platform is a hackathon-built web app featuring an AI model for fire detection via IP camera and a weather report API to boost situational awareness. Led by a team of six, it achieved over 95% fire detection accuracy and plans to expand with alert systems and rainfall prediction.
HTML
CSS
JavaScript
Bootstrap
Python
Flask
OpenCV
Tensorflow
Keras
HTML
CSS
JavaScript
Bootstrap
Python
Flask
OpenCV
Tensorflow
Keras
Self Driving Car (JS)
HTML
CSS
JavaScript
Self Driving Car (JS)
Self Driving Car (JS) is a browser-based simulator that uses JavaScript to model autonomous vehicle behavior. Users can train the car by saving sensor parameters, tweak simulation settings like sensor count and car numbers, and visualize learning iterations live in the browser.
HTML
CSS
JavaScript
Reinforcement Learning
HTML
CSS
JavaScript
Reinforcement Learning
ChatWithWebsite
Next.js
TailwindCSS
Upstash Redis
ChatWithWebsite
ChatWithWebsite is a web app that lets users interact with any website by entering its URL into a message box, enabling quick redirection and information retrieval. It features a clean UI, seamless navigation, and responsive design for all devices.
Next.js
TailwindCSS
Upstash Redis
Rag Chat
TypeScript
Next.js
TailwindCSS
Upstash Redis
Rag Chat
TypeScript
VisionMATE: Assistive Technology for Visually Impaired
React.js
TensorFlow.js
Web Speech API
VisionMATE: Assistive Technology for Visually Impaired
VisionMATE is a full-featured web app designed to support visually impaired students through real-time object detection, text-to-speech, speech-to-text, and voice-controlled task management. Built with React.js, TensorFlow.js, and Node.js, it offers a responsive, accessible interface powered by motion and speech technologies.
React.js
TensorFlow.js
Web Speech API
TailwindCSS
Node.js
Express
MongoDB
JWT Auth
Python
Yolo
OpenCV
React.js
TensorFlow.js
Web Speech API
TailwindCSS
Node.js
Express
MongoDB
JWT Auth
Python
Yolo
OpenCV
Two-Wheeler Road Surface Classifier
Python
MIT App Inventer
LSTM
Two-Wheeler Road Surface Classifier
An Android app that records accelerometer and gyroscope data, builds a dataset, and trains an LSTM to classify 3‑second samples into Kankar, Bitumen, Concrete, single speed breaker, and multiple speed breakers. It enables on-bike, real-time surface classification and can be extended to rash-driving alerts using the same motion dataset.
Python
MIT App Inventer
LSTM
Pandas
Numpy
TensorFlow
Python
MIT App Inventer
LSTM
Pandas
Numpy
TensorFlow
QR Code Generator
HTML
CSS
JavaScript
QR Code Generator
This is a simple QR Code Generator built using HTML, CSS, and JavaScript. It allows users to generate QR codes from URLs and download them as PNG images with customizable sizes.
HTML
CSS
JavaScript
HTML
CSS
JavaScript
Logistic Regression Projects
Python
Numpy
Pandas
Logistic Regression Projects
These are some practice Logistic Regression models [Insurance_dataset.ipynb, Load_Digits_Dataset.ipynb, Iris_Flower_Dataset.ipynb, HR_Dataset.ipynb]
Python
Numpy
Pandas
Matplotlib
Scikit-Learn
Python
Numpy
Pandas
Matplotlib
Scikit-Learn
MNIST Number Recognition
Python
Numpy
Pandas
MNIST Number Recognition
Simple MNIST Neural Network from scratch In this notebook, I implemented a simple two-layer neural network and trained it on the MNIST digit recognizer dataset. It's meant to be an instructional example, through which you can understand the underlying math of neural networks better.
Python
Numpy
Pandas
Matplotlib
Python
Numpy
Pandas
Matplotlib
AiPlaneTech Official Website
Next.js
TypeScript
TailwindCSS
AiPlaneTech Official Website
This is the Official website of AiPlaneTech Pvt Ltd India. An AI based startup, Empowering AI & ML Solutions Bringing cutting-edge machine learning models to organizations in need.
Next.js
TypeScript
TailwindCSS
GCP
Next.js
TypeScript
TailwindCSS
GCP
React Jobs Project
React.js
JSON Server
MongoDB
React Jobs Project
This project is a job listing website built with React.js, inspired by a YouTube tutorial by Traversy Media. It showcases modern React practices, including functional components, hooks, and reusable design.
React.js
JSON Server
MongoDB
Express
React.js
JSON Server
MongoDB
Express
Personal Portfolio
NextJs
MongoDB
Vercel
Personal Portfolio
This is a completely data-driven website that fetches data from MongoDB and images from Cloudinary and has an admin panel to control all this data.
NextJs
MongoDB
Vercel
Cloudinary
NextJs
MongoDB
Vercel
Cloudinary

Experience & Certifications

My professional journey, internships, and certifications that have shaped my expertise.

Work Experience

AI Architect & Project Lead Intern
AlPlaneTech
Intern
February 2025 - July 2025

Led and conducted two hands-on AI workshops, training 20+ participants on AI, ML, NLP, and Image Processing fundamentals.

Key Achievements:

  • Architected and deployed full-stack solutions using Google Cloud Platform (GCP)
  • Set up scalable infrastructure, DevOps pipelines, and cloud storage services
  • Spearheaded multiple AI-driven projects including RAG System and VQA System
  • Trained 20+ participants in AI/ML fundamentals through hands-on workshops

Technologies:

Python
GCP
NLP
Computer Vision
DevOps
Python
GCP
NLP
Computer Vision
DevOps
R&D Intern - Robotics
RoboAi Hub
Intern
June 2024 - August 2024

Developed advanced robotics projects including line followers, robotic hands, and Bluetooth-controlled multifunctional cars.

Key Achievements:

  • Developed line follower robots with advanced sensor integration.
  • Built robotic hands with precise motor control systems.
  • Created Bluetooth-controlled multifunctional cars with IoT integration.
  • Gained expertise in IoT, wireless communication, and PCB design.

Technologies:

ESP32
Arduino
Raspberry Pi
IoT
PCB Design
Automation
TinkerCAD
C++
Python
ESP32
Arduino
Raspberry Pi
IoT
PCB Design
Automation
TinkerCAD
C++
Python
Artificial Intelligence & Machine Learning Intern
Nexus-Info
Intern
May 2024 - July 2024

Built a Q&A chatbot for college admissions with contextual understanding and real-time backend integration.

Key Achievements:

  • Built Q&A chatbot for college admissions with contextual understanding.
  • Implemented real-time backend integration for seamless user experience.
  • Developed disease prediction system using AI&ML techniques.
  • Focused on data preprocessing, feature selection, and model optimization

Technologies:

Python
Machine Learning
NLP
Data Preprocessing
Model Optimization
Model Creation
Python
Machine Learning
NLP
Data Preprocessing
Model Optimization
Model Creation
Summer Robotics Workshop
ESRC MBM
Robotics Trainee
June 2023 - August 2023

ESRC Workshop 2023 was an intensive robotics training program where participants gained hands-on experience with Arduino, ESP32, and Raspberry Pi boards. Over the course of the workshop, Harshit led a team of five to build multiple robotics projects including RC cars, line followers, robotic hands, Bluetooth-controlled vehicles, and IoT-enabled systems. The workshop emphasized practical learning, culminating in a Python-based IoT device with OpenCV for object detection. Harshit was awarded a Certificate of Excellence for outstanding performance and full attendance.

Key Achievements:

  • Led a team of 5 members to successfully complete multiple robotics projects, including RC cars, line followers, robotic hands, Bluetooth-controlled cars, automated LED’s and IoT-enabled devices.
  • Gained hands-on experience with programmable boards such as Arduino, ESP32, and Raspberry Pi.
  • Developed a Python-based IoT device using OpenCV for object detection.
  • Awarded Certificate of Excellence for outstanding performance and achieving 100% attendance in the workshop.

Technologies:

Arduino
PCB Design
IoT
C++
Python
OpenCV
ESP32
Raspberry Pi
Arduino
PCB Design
IoT
C++
Python
OpenCV
ESP32
Raspberry Pi

Certifications

FREE OpenCV Bootcamp
OpenCV University
January 2025

In this Journey, I started with image basics and then moved on to manipulating images by cropping, resizing, and flipping. I learned to annotate and enhance images using mathematical operations. I accessed the camera, recorded videos, and applied image filtering. I explored image alignment, creating panoramas and HDR images. Finally, I delved into object tracking, face detection, TensorFlow object detection, and OpenPose for human pose estimation.

Skills Gained:

Computer Vision
OpenCV
Image Processing
Python
Computer Vision
OpenCV
Image Processing
Python
100 Days of Code: The Complete Python Pro Bootcamp
Udemy
November 2024

100 Days of Python Code – Complete Python Pro Bootcamp is a hands-on coding journey through Dr. Angela Yu’s Udemy course, where Harshit completed 100+ real-world projects across automation, web development, data science, and game design. The repository documents daily progress, showcasing skills in Python, Flask, Selenium, Pandas, Tkinter, and more—culminating in a rich portfolio of deployable apps and intelligent tools.

Skills Gained:

Python
PyCharm
JupyterNotebook
GoogleColab
Flask
HTML5
CSS3
BootStrap4
Tkinter
Turtle
Selenium
BeautifulSoup
Pandas
Numpy
Matplotlib
Plotly
Seaborn
Scikit-Learn
SQLite
PostgreSQL
Github
Heroku
Gunicorn
Git
Python
PyCharm
JupyterNotebook
GoogleColab
Flask
HTML5
CSS3
BootStrap4
Tkinter
Turtle
Selenium
BeautifulSoup
Pandas
Numpy
Matplotlib
Plotly
Seaborn
Scikit-Learn
SQLite
PostgreSQL
Github
Heroku
Gunicorn
Git
Letter of Participation
E-Yantra Robotics, IIT Bombay
March 2024

I participated in the E-Yantra Robotics Competition at IIT Bombay under the theme GeoGuide, where our team built "Vanguard"—a robotic vehicle powered by machine learning, image processing, and georeferencing. This hands-on experience sharpened our skills in GIS, remote sensing, and wireless communication, earning us a certificate of participation.

Skills Gained:

Leadership
Machine Learning
Image Processing
GIS
Remote Sensing
Leadership
Machine Learning
Image Processing
GIS
Remote Sensing
Certificate of Appreciation
GDSC MBM
January 2024

I was recognized as one of the top performers at the Google Summer of Code Primer Workshop hosted by GDSC MBM! This experience deepened my understanding of real-world contributions and sharpened my skills in Git and GitHub, fueling my passion for impactful tech innovation.

Skills Gained:

Git
Github
Python
Git
Github
Python
Graphics Coordinator
TechKriti MBM
September 2023

I received a Certificate of Coordination for my role in TechKriti 2023 at MBM University, Jodhpur! This experience sharpened my skills in Graphics and Management while celebrating teamwork, creativity, and leadership. Grateful for the journey and excited for future milestones!

Skills Gained:

Leadership
Team Management
Graphics
Canva
Figma
Leadership
Team Management
Graphics
Canva
Figma
Certificate of Excellence - Embedded Systems & Robotics Club
ESRC MBM
August 2023

Received Certificate of Excellence for 100% attendance and performance in the ESRC Summer Workshop 2023 at MBM University, under Prof. Alok Singh Gehlot and Prof. Abhishek Gaur, and gained hands-on experience in robotics and embedded systems through projects like a line follower robot, Bluetooth car, Tic-Tac-Toe robot, ESP32 with camera, and more. I am grateful to all my mentors for their guidance.

Skills Gained:

Arduino
PCB Design
IoT
C++
Python
OpenCV
ESP32
Raspberry Pi
Arduino
PCB Design
IoT
C++
Python
OpenCV
ESP32
Raspberry Pi

Get In Touch

Have a project in mind or want to collaborate? I'd love to hear from you. Let's discuss how we can work together to bring your ideas to life.

Let's Connect

I'm always interested in new opportunities, collaborations, and interesting projects. Whether you have a question or just want to say hi, feel free to reach out!

Email

harshitsoni11664@gmail.com

Send me an email anytime

Location

Jodhpur, Rajasthan, India

Available for remote work

Send me a message
Fill out the form below and I'll get back to you as soon as possible.