Hi I am Dev Haral 
,
I am a recent B.Tech graduate in Mechanical Engineering with a minor in Data Science from IIT Mandi. My burning passion for the exciting and ever-evolving fields of Deep Learning and Natural Language Processing, combined with my engineering background, gives me a unique interdisciplinary approach to problem-solving. This blend of skills opens up a world of possibilities for me. I am excited to embark on this journey of growth and development, eagerly looking forward to the opportunities and challenges that lie ahead.
Python
90%
Tensorflow
50%
Pytorch
50%
C++
80%
HTML/CSS
70%
React
70%
Machine Learning
80%
Deep Learning
80%
Reinforcement Learning
50%
NLP
70%
Data Structure and Algorithms
50%
Nestled in the Sivalik Range of the Himalayas, away from the bustle of the metropolis, a new abode of learning has germinated. A few hours before the Himalayan resort Kullu in Himachal Pradesh, IIT Mandi has been established with the vision to be a leader in science and technology education, knowledge creation and innovation, in an India marching towards a just, inclusive and sustainable society.
Working with Data Science team.
Developed a resume parsing system, extracting information from PDFs and generating embeddings for efficient storage and analysis. Established and managed a Neo4j database for storing diverse datasets, optimizing for efficient querying and relationships.Conducted Twitter scraping for tweets and replies, storing the data for sentiment analysis and trend identification. Implemented OpenAI’s API for a PDF-based question-answering system, enhancing information retrieval from resumes.
Maintained and optimized pretrained Variational Autoencoder (VAE) models for generating music, which were built on the TensorFlow framework on strong platform Google Colab Pro. Built an API to connect Reaper software, a digital audio workstation, with the VAE models, allowing for seamless integration and music generation through a specific Pipeline Developed a recommendation system for music based on genres,utilizing deep learning techniques such as LSTM and leveraging the output of the VAE models.