About

Mohit Appari

Student

Florida State University

I'm an aspiring Data Scientist driven by an insatiable thirst for knowledge and a passion for uncovering insights hidden within the vast expanse of data. Currently, I am focused on building robust data pipelines, and machine learning.

Passionately engaged in the realms of machine learning and data engineering, I'm constantly immersed in diverse projects aimed at pushing boundaries and driving innovation. As a staunch advocate for open source collaboration, I actively seek opportunities to work alongside new collaborators. Currently, my focus is on building robust data engineering pipelines and leveraging cutting-edge techniques in LLMs and artificial intelligence to tackle real-world challenges.

If you're embarking on an exciting project or seeking fresh perspectives, don't hesitate to reach out via Mail or Whatsapp. I'm always eager to connect, exchange ideas, and explore new avenues of exploration and growth together.

About this site

Welcome to my home on the internet. This site functions as a blog/portfolio, a place to share code and thoughts. Opinions of my own.

I learnt how to build this site from the most awesome people in the community:

Timothy's Next.js and Tailwind CSS template: Template starter where I bootstrapped the project.
Einar Guðjónsson: Now page, navigation style, animations and much more.

Experience

Research Assistant - Artificial Intelligence @ Florida State University

June 2024 – Present

‣

Researched and experimented with GANs, diffusion models, UniMat by Google, and CDVAE by Microsoft, exploring novel architectures for material synthesis.

‣

Implemented NequIP on a dataset of 11,043 materials, achieving 95% accuracy with customized YAML configurations, including 4 rotational invariant layers and equivariance in irreps, optimized with AMSGrad and Adam.

‣

Fine-tuned hyperparameters such as learning rate (0.005) and batch size (100), improving model efficiency by 20% with a focus on energy prediction.

~~~

Associate Data Engineer @ S&P Global

January 2022 – July 2023

‣

Automated data ingestion processes for over 50,000 files, including PDFs, images, and TIFFs, using advanced OCR and computer vision tools, achieving a 90% accuracy rate in text extraction.

‣

Developed Python pipelines to extract, transform, and load (ETL) data into S3 buckets and servers, utilizing regex and Grooper for complex text extraction, improving overall efficiency and speed by 90%.

‣

Deployed executable pipelines to streamline data ingestion, leading to seamless integration of dashboards on Power BI.

~~~

Machine Learning Intern @ LG

June 2021 – August 2021

‣

Developed a drowsiness detection application using WebOS and OpenCV on Raspberry Pi 4, improving driver safety by 50%.

‣

Secured 3rd place in an Ideathon by collaborating on a smart logistics system using computer vision.

‣

Collaborated with a team of 4 to develop logistic solutions using computer vision and machine learning.

~~~

Software Development Intern @ ITC

January 2021 – March 2021

‣

Developed a Node.js web application using Express.js and MySQL for a master data management system, improving user experience and feasibility by 50% across 10+ categories.

‣

Implemented Google authentication and utilized bcrypt for password encoding, enhancing security and streamlining the ticket creation process.

‣

Designed and deployed a RESTful API for CRUD operations, enabling seamless data retrieval and modification.

~~~