Pranjal Soni Email and Phone Number

• Built and implemented an automated demand forecasting engine for FMCG products, streamlining the data science workflow by automating feature generation, data transformation, and model building. Achieved a 33% reduction in runtime, enhancing operational efficiency. Demonstrated a remarkable 20% improvement in forecasting accuracy.• Engineered a range-based forecasting model utilizing residual bootstrapping techniques to mitigate bias, resulting in enhanced forecast accuracy.• Contributed to developing key tools, including a classification engine for seamless model development, training, and testing; an EDA app enabling code-free exploratory data analysis with exportable reports; and a cookie-cutter setup to streamline processes for the data science team.• Designed and implemented a multi-functional audio analysis tool with Azure services & OpenAI´s LLM models, including features for speech-to-text transcription, language translation, audio summarization, sentiment analysis, and a chatbot for interactive engagement with audio content.• Developed multiple Streamlit dashboards to facilitate model delivery to end users, ensuring efficient monitoring and eliminating dependencies on Excel.• Led data cleaning and exploratory data analysis initiatives, specializing in classical time series models such as ARIMA, SARIMA, and Prophet. Also, I proficiently worked with machine learning and deep learning algorithms, including Random Forest, XGBoost, LightGBM, Neural Prophet, and LSTM.

• Worked on digital signals, filtering, analyzing, and preparing the data. Applied various statistical methods to generate insights and model the signals.• Setup an end-to-end automated architecture for the organization for processing and loading the data into the data warehouse for analytics purposes. Also, set up analytical infrastructure for delivering the analytics using PowerBI. • Worked with various teams i.e. Data Science, Product, Sales, Finance, and Marketing to understand and deliver their analytics requirements.• Perform data visualization for time series data using Seaborn and Matplotlib. And also explicitly worked on Pandas, Numpy, Bigquery, Linux, and Google Console.

• Built deep learning models for computer vision project using thePytorch framework for recognizing text from the images • Worked with transfer learning technique and used pre-trainedmodels like Resnet, Efficientnet, Nfnets, and Attention models• Extract business keywords from articles using collocations and theRAKE algorithm• Performed data collection, preprocessing, and labeling. Improverunning time of the python scripts to enhance the efficiency of theprograms