I am a recent Computer Science (Data Science minor) graduate from Northern Kentucky University with 6+ years of coding experience. Skilled in Software Engineering, Machine Learning, and Data Science, with hands-on experience in full-stack development, building scalable data pipelines, and implementing end-to-end ML solutions.
I’ve built Full stack Projects like a custom Job Search Platform and also worked on Machine Learning projects Sentiment Analysis for 20M+ reviews, fine-tuning models like DistilBERT, and Designing ML pipelines that improved defect detection by 30%. I have also worked on deploying Microservice architecture on Azure and AWS.
Passionate about building user-centric applications, solving challenging data problems, and creating innovative platforms that drive impact through code. Always ready to learn, optimize, and deliver results.
Email: Sebikash10@gmail.com
Live Link | Appjob.net
Technologies: Next.js, PostgreSQL, TypeScript, Express.js, Docker, Azure, Clerk.dev Authentication, python (web-scraping)
This is a full stack application that lets user get the most matching jobs for their profile. This application is built on a microservice architecture i.e frontend (Next.js), backend(Express), database(PostgreSQL), llama3.1 model, web scraper(python) and is deployed on azure.
-Web Scraper is a cron job and runs every hour and uploads the new jobs on the database.
-The llama3.1 also runs every minutes and vectorizes the job description and stores it in to the database and does the same for the resume uploaded by the user.
-Users sign in using gmail account.
-Authentication is handled by clerk.
-The app also has a filter option where users can search jobs using job titles, filter remote jobs and search for internships/contract/fulltime or part time jobs.
Live Link
Technologies: JavaScript, WebSockets, MongoDB
This is a Multiplayer online real-time BINGO Game. Player can join or host the game; it includes real time chat functionality with web sockets.
Live Link
Technologies: Python, Convolutional Neural Networks, TensorFlow, Keras, Flask, Azure
This app classifies images into predefined categories by training a Convolutional Neural Networks (CNNs) model. It provides real-time predictions with a user-friendly interface for uploading and classifying images. It is hosted on Azure.
Github Link
Technologies: Java, Socket Programming, Multithreading, Client- Server Communication
This project implements a multithreaded client-server application to play the game Mastermind. The server generates a random 4-color code, and the client attempts to guess the sequence within 20 attempts. The server provides feedback on the correctness of each guess.
Github Link
Technologies: React, Tailwind CSS, OAuth
Designed and developed a university collaboration platform connecting researchers and students in a group of 2. Integrated advanced filtering and recommendation systems for real-time collaboration
Github Repo
Technologies: Kafka, Spark, YOLOv3, OpenCV
Created a video object detection pipeline, achieving 85% detection accuracy. Optimized video frame processing to reduce latency in real-time applications.
Google Collabs Docs
Technologies: DistilBERT, PySpark, Natural Language Toolkit (NLTK), Keras, Transformers, Torch
This project applies sentiment analysis to RateMyProfessor reviews using NLP techniques. It preprocesses data with Pandas, Spark on local machine, NumPy and uses pretrained Transformer models with PyTorch for fine-tuned sentiment classification. TextBlob provides initial sentiment scores, while transfer learning improves model accuracy, capturing nuanced feedback.
Analyzed over 20M+ Rate My Professor reviews with 86% sentiment accuracy.
Technologies: TensorFlow, Keras, Flask, PostgreSQL
Developed an ML pipeline for industrial defect coil production detection for North American Stainless , increasing detection accuracy by 30%. Preprocessed 50GB+ of data and optimized training for scalable deployment.
Excited to work on impactful projects and collaborate with other tech enthusiasts! 🎯
