Sarthak Kalia

Sarthak Kumar Kalia

AI/ML Engineer

sarthakkalia45@gmail.com

Welcome to My Portfolio!

I'm Sarthak Kumar Kalia, a dedicated Data Scientist with a focus on Artificial Intelligence and Machine Learning. My passion lies in developing robust algorithms, uncovering actionable insights from vast datasets, and creating accurate, scalable machine learning models. My journey has been marked by active participation in hackathons, collaboration on cutting-edge projects, and a relentless drive to push the limits of data science.

I am particularly fascinated by the transformative power of AI and its potential to redefine industries. Explore my work, and feel free to connect with me!

Download résumé
Interests
  • Research
  • Algorithms
  • Generative AI
  • Natural Language Processing
  • Machine Learning
Education

  • B.Tech in CSE (Artificial Intelligence and Machine Learning),

    2022-2026

    SUIIT (Sambalpur University Institute of Information Technology)

Experience

 
 
 
 
 
Predulive
NLP and Text Analytics (Machine Learning) Internship
Predulive Labs
Nov 2024 – Present Hybrid
  • Developed AI-based models for Drone Technology Integration, enhancing drone performance and capability.
  • Preprocessed and analyzed text data to extract sentiment insights, enabling informed decision-making.
 
 
 
 
 
GDG
AI/ML Lead
Google Developer Groups (GDG) on campus SUIIT
2024 – Present On-Site
  • Spearhead campus-wide initiatives focused on artificial intelligence and machine learning, driving innovation through workshops, hackathons, and collaborative projects.
  • Leadership & Technical Experience
  • Community Engagement & Collaboration
 
 
 
 
 
hp
AI/ML Engineer Intern
NextGen AI Frameworks Pvt. Ltd.
Sep 2024 – Nov 2024 Remote
  • Designed, implemented, and trained AI/ML models to address specific business problems, delivering tailored solutions for diverse use cases.
  • Conducted iterative testing and optimization to improve model performance and scalability.
  • Collaborated with cross-functional teams to integrate AI-driven insights into business workflows.
 
 
 
 
 
Winner List
Google
VirtuEye - Virtual Eye for Visually Impaired Person
Dec 2023 – Jun 2024
  • Developed a real-time haptic feedback system for visually impaired users, integrating machine learning-based obstacle detection with a fine-tuned CNN model (MobileNetV3), ensuring faster and accurate object detection on edge devices
  • Implemented YOLO with OpenCV for real-time object detection and applied OCR for text extraction, enhancing navigation efficiency and environmental awareness for users.
  • Created and annotated a custom roadside dataset using RoboFlow, implementing advanced data preprocessing techniques to improve model accuracy in diverse, real-world conditions.
  • Utilized HuggingFace NLP models to generate descriptive audio output from detected objects and text, providing users with a seamless and accessible auditory experience.
  • Optimized system performance for scalability, ensuring the solution can be deployed for mass accessibility, and collaborated with stakeholders to enhance usability based on user feedback.

Skills

AI & Data Science

1+ Years XP

Machine Learning

Deep Learning

Data Visualization

Predictive Modeling

Statistical Analysis

Computer Vision

Natural Language Processing (NLP)

Programming

2+ Years XP

Python

Java

C

SQL

NoSQL

Cloud Computing

1+ Years XP

Google Cloud Platform

AWS (Amazon Web Services)

Docker

CI/CD Pipelines

Front End

1+ Years XP

HTML

CSS

Figma

Canva

Back End

1+ Years XP

Python - Flask, FastAPI, RestAPI

MLOps & Deployment

1+ Years XP

ML Lifecycle Management

Model Deployment

Serving APIs (FastAPI, Flask)

Cloud-Based ML (Google Cloud AI, AWS SageMaker)

Miscellaneous

2+ Years XP

Git & GitHub

Linux

DevOps Practices

Software Development

Accomplish­ments

ISRO
AI/ML for Geodata Analysis
ISRO Sept 2024
  • Earned a certificate from the Indian Space Research Organisation (ISRO) for completing specialized training in September 2024.
  • Gained hands-on experience in space-related technology applications through ISRO's verified program.
See certificate
Coursera
Data Science Masters 2.0
PW Skills Aug 2024
  • Completed the PW Data Science Masters 2.0 course, where I honed my expertise in machine learning, data analysis, and scalable model development. The course provided extensive hands-on experience with real-world projects, allowing me to apply advanced techniques in a practical setting.
  • I gained proficiency in building end-to-end data pipelines, optimizing models, and deriving actionable insights from complex datasets. This experience further solidified my ability to create impactful, data-driven solutions.
See certificate
udemy
Building GenAI App & Hands-on Projects
Udemy Aug 2024
  • Gained experience in building GenAI applications and hands-on projects using Large Language Models (LLMs). Applied cutting-edge techniques to develop AI-driven solutions that leverage advanced natural language processing capabilities.
See certificate
Google
Google Solution Challenge - Top 100 Finalist
Google May 2024
  • Selected as a Top 100 Finalist in the Google Solution Challenge, recognized for developing innovative solutions to real-world problems.
  • An AI-powered navigation system designed to assist visually impaired individuals by providing real-time feedback about their surroundings through haptic and audio cues.
See certificate

Projects

*
All Machine Learning Deep Learning Computer Vision NLP/LLM
Emergency Preparedness Advisor bot
For this project, I fine-tuned the GPT-2 model to develop an AI-powered emergency preparedness chatbot. I performed extensive text preprocessing, which included cleaning, tokenization, and formatting the dataset to ensure it was suitable for model training. I then fine-tuned the GPT-2 model on the curated dataset, focusing on enhancing its ability to generate context-aware and relevant responses. This process enabled the bot to effectively provide tailored emergency preparedness advice based on user inputs, improving its accuracy and performance in real-world applications.
Sarthak Kalia
GitHub
HerbWise: AI-Based Herb Classification and Chatbot System
Developed a comprehensive AI-driven solution that combines a fine-tuned Convolutional Neural Network (CNN) for herb classification with an interactive chatbot. Leveraged MobileNetV2 and VGG16 models, selecting MobileNetV2 for its superior accuracy. The chatbot, powered by natural language processing (NLP), provides users with instant information about the uses and benefits of various herbs, enhancing user engagement and learning.
Sarthak Kalia
HerbWise
VirtuEye - AI-Smart Glass for Visually Impaired People
VirtuEye is an innovative AI-powered smart glass designed to assist visually impaired individuals in navigating their environments safely and independently. The system employs a fine-tuned MobileNetV3 Convolutional Neural Network (CNN) for real-time object detection, integrated with YOLO (You Only Look Once) for immediate identification of obstacles. Optical Character Recognition (OCR) technology is utilized to extract text from images, providing users with vital information about their surroundings. Additionally, a natural language processing (NLP) model enhances user interaction by converting detected objects and text into meaningful audio descriptions. A comprehensive dataset was created and annotated using RoboFlow, ensuring high-quality data for improved model performance and accuracy.
Sarthak Kalia with his team
VirtuEye Result YouTube
VirtuEye
Spam/Ham Classifier
Developed a robust Spam/Ham Classifier utilizing natural language processing (NLP) techniques, including Bag of Words (BoW) and TF-IDF for effective message categorization. Implemented the Multinomial Naive Bayes algorithm to enhance classification accuracy, leveraging advanced text processing methods and feature extraction techniques to optimize performance.
Sarthak Kalia
Repository
Spam-Ham
CNN- Architecure
Contributed to a GitHub repository focused on image classification tasks by developing a Convolutional Neural Network (CNN) architecture. This includes a comprehensive video notebook and a paper to facilitate learning about advanced deep learning techniques, showcasing hands-on experience with state-of-the-art neural network methodologies.
Sarthak Kalia with his team
Repository Content
CNN
Invoice Extractor LLM app
Developed an Invoice Extractor LLM application that leverages large language models (LLMs) to automate the extraction of key information from invoices. Implemented advanced text processing and entity recognition techniques to streamline data extraction and improve accuracy in financial document management.
Sarthak Kalia
Repository
Diamond Price Prediction
Developed a predictive model for estimating diamond prices using regression analysis and implemented a machine learning pipeline that automates the prediction process. Evaluated various models, including Linear Regression, Lasso, Ridge, ElasticNet, and Decision Trees, through this pipeline. Conducted thorough feature engineering and data preprocessing to optimize model performance, ultimately selecting the best-performing model based on accuracy metrics. This project showcases proficiency in data analysis, model selection, and the application of machine learning techniques to solve real-world problems.
Sarthak Kalia
Repository
Diamond Price Prediction
Shopping Management System
Developed an online shopping management system using Java, implementing Object-Oriented Programming (OOP) principles such as abstraction, inheritance, polymorphism, and encapsulation. Integrated JDBC for seamless database interaction with MySQL, allowing for efficient management of user and product information. Key features include user registration and login, product browsing, shopping cart management, and payment handling. The system is structured with a clear modular design, promoting maintainability and scalability.
Sarthak Kalia
Repository
Shopping Management System

Contact Us

Feel free to reach out to us using the contact details below: