Engineer with a passion for building. Data Engineer at FINRA

Nithin Sameer Yerramilli  • Peerlist

• It is an attempt to try and make a community out of the independent filmmaking industry.It is a website used to connect one filmmaker with another regardless of commonality based solely on the skill and merit of each individual. It helps establish connections that prove to be useful and worthwhileIt is also a platform that is used to provide information about the various tools available to an individual foraying into the field of filmmaking. It provides them with a list of all the various resources they may need in every possible price range. • It uses a combination of Django for the front end and Python and SQLite 3 for the back end. It displays an ad posted by someone to prospective connects and provides information about the field and the tools involved from a mere novice to a professional.

Cinescope

Achieved 93.15% accuracy in classifying Sloan Survey celestial objects by optimizing Logistic Regression, K-Means, and Decision Trees in Databricks/Python/Keras, validating redshift’s significance via EDA and Deep Learning.

Classifying Celestial Objects with Spectral Data from the Sl

• Developed PollutionPulse web app to predict Mumbai’s air quality index (AQI) in real-time, achieving MAPE consistently under 20%, with enhancements driving it below 10% for validation data. • Incorporated an hourly updating responsive feedback loop, enhancing prediction accuracy by reducing model error to 13-15%, showcasing strong short-term forecasting performance.

Forecasting Air Quality Index for Mumbai

• Identified age at release, gang affiliation, and employment status as significant predictors of recidivism with 76.5% accuracy using R-based Stepwise Selection and Chi-Square analysis in logistic regression. • Implemented Decision Tree analysis using R, identifying critical employment thresholds that differentiate recidivism probabilities, with employment status influencing recidivism predictions at a key threshold of 64.3% days employed.

Predicting Recidivism Based on Demographical Bias and Crimin

Engineered a robust recipe recommendation system using TF-IDF vectorization and OpenAI’s GPT-3.5-turbo model, delivering personalized and culturally adapted recipe suggestions through an intuitive Streamlit frontend.

RecipeMate

• It is a platform using which a little uniformity can be brought to the whole recruitment process that occurs in colleges during the final two semesters. • It tries to bring awareness to the students about the various options that are available to them and also to test their all-round development. It also acts as a valuable system for companies by shortening the application list and reducing their workload. • It uses a combination of Django for the front end and Python and SQLite 3 for the back end. It provides a consolidated list of students to the companies who have passed their criteria and a consolidated companies list to the students to give them a better idea of the available options.