Satyam Verma

I am a senior at MMDU studying Computer Science & Engineering. Over the past year, I have been working on improving my data structure and algorithm concepts, side-by-side with learning the concepts of Transformers and open-source LLMs.

I previously won The Smart India Hackathon 2022, where I worked on a machine learning model to evaluate handwriting and grade it.

I'm passionate about solving critical challenges in Machine Learning. I'm particularly interested in leveraging underlying data and problems to build more robust generalizable models; deep learning models that utilize structural dependencies will be critical to achieving generalization.

Email  /  Resume  /  Github  /  LinkedIn  /  Twitter  /  Leetcode

profile photo
News

July. '23  

 Selected in Singapore India Hackathon under Startup Track.

Nov. '22  

 Selected in UNESCO India Africa Hackathon.

Aug. '22  

 Won the Smart India Hackathon.
Projects
UnScrawl (June. 2022 - Aug. 2022)
Handwriting Assessment App | Smart India Hackathon. Source.
  • Developed a deep learning-based application using Python, TensorFlow, and OpenCV that captures handwritten pages and highlights wrongly formed alphabets to enhance handwriting of students.
  • Created data sets, implemented the deep learning model, and designed image processing and analysis algorithms.
  • Developed a Flask API and set up CI/CD pipelines for Dockerized Azure Containers, ensuring smooth deployment and scalability.
WaterSense (Nov. 2022)
Soil Moisture Measuring App. Source.
  • Created a machine learning model using TensorFlow, OpenCV, and Python, with a locally generated data set, to build an app that uses a standard image from a mobile camera to determine soil moisture..
  • Forecast soil moisture and determined the best time to irrigate using weather information and soil type.
TalkHealthy (Apr. 2023)
AI-Powered Fitness Chatbot. Source.
  • Developed a web application using Python that provides users with personalized answers and guidance on diet, gym, and exercise-related questions.
  • Implemented using the Llama model to facilitate natural language processing and understanding of user queries.
  • Utilized a vector database to store a vast amount of data on diet, nutrition, fitness, and exercise to integrate a semantic query system that combines the user queries with relevant contextual information, enhancing the accuracy and relevance of responses.

Updated August 2023.

Design and source code forked from Jon Barron's website.