Hey there!
I'm Aadvait Hirde
I am a Computer Science and Business student with an avid interest in Machine Learning and Full Stack Web Development. My curiosity drives me to explore the intricacies of technology and its endless possibilities. I thrive on challenges and am always eager to learn and grow, both professionally and personally.
About Me
My journey in tech has been marked by a diverse set of experiences. I have ventured into the innovative world of Augmented Reality, specifically focusing on indoor navigation projects. I've also published a research paper on using transformers and deep learning with stance detection to forecast cryptocurrency price movement. In terms of raw coding, I possess a strong foundational knowledge in programming languages such as Flutter, Dart, Java, Python, C++, and I've honed my skills in web development with HTML, CSS, JavaScript and the MERN stack alongside machine learning. As a future tech professional, I am excited about the opportunity to merge my technical abilities with my business acumen to make an impact in the field. I am always on the lookout for collaborative projects, research prospects, internships, or conversations that can broaden my understanding and experience in this ever-evolving industry.
My Experience
President of the Machine Learning in Econometrics Club
Working on leading a team of 50+ students in exploring cutting-edge applications of machine learning algorithms to economic modeling, organizing weekly workshops, hosting industry speakers, and spearheading a successful annual hackathon that aims to attract 200+ participants from top universities.
Software Development Intern at MyCaptain
Developed MERN stack web app boosting user engagement 15% for 50,000 daily users. Built an ML model with Python/scikit-learn for targeted marketing and implemented RESTful API using Express.js, MongoDB for integrations. Created Python data pipeline using Pandas and NumPy to increase efficiency by 40%.
My Projects
The Locus App
Aug 2022 - Nov 2022Locus is an AR-based indoor navigation app. I implemented C# scripts for AR interactions, object placement, and UI management. The project integrated AR Foundation, ARCore, and ARKit modules for Android and iOS platforms, leveraging Depth APIs for accurate AR projections. A QR code-based tracking system was designed, where each code corresponded to a specific location, updating the user's position in the Unity model upon scanning. Unity's scene management was utilized to create a scale model of the campus, assigning navigation targets to each classroom. Using this, real-time indoor pathfinding was enabled by rendering AR and minimap lines from the user's location to the destination.
Read More on Khaleej Times
BLAST AI Research Paper
Jun 2022 - Nov 2022I developed a stance detection model using the RoBERTa transformer for analyzing tweet sentiment towards Bitcoin, achieving an 80% accuracy rate. Additionally, I implemented an LSTM-based Recurrent Neural Network (RNN) for Bitcoin price forecasting, with a mean absolute error of $1144.85 over a 5-day span. For sentiment analysis, logistic regression was applied, yielding a 78% accuracy rate. Sklearn was used for data preprocessing and model evaluation, while the Huggingface library facilitated access to pre-trained transformer models. In training the RNN, the Adam optimizer was employed, and suggestions were made for future hyperparameter tuning using GridSearchCV and Bayesian optimization techniques.
Read More on IEEE