Atharva Joshi

Designed python scripts to generate 55 to 60 daily, weekly and end-of-period reports on to be scheduled basis at their designated times; enabling customer’s business board to get weekly stats for their business. Innovatively simplified the transfer mechanisms of larger reports to reduce their transfer times by 30% to 40% using SFTP transfer mechanisms, which accelerated and streamlined the delivery process of scheduled reports. Streamlined the monitoring protocol of daily scheduled processes by automating manual monitoring tasks. Drastically reduced occurrences of process failures by introducing proactive failure resolution mechanism in the monitoring which resulted in a smooth and error-free execution of daily processes.

Worked on development of various machine learning products using frameworks like RASA for NLP based Chatbot. Independently developed ML Interpretability Module for the Generic Machine Learning Pipeline (GMLP) with requiring extensive research in interpretability and explainability concepts like shapely values, partial dependence, etc. which gave the GMLP the feature of explaining the predictions. Built a demand forecasting engine to forecast demand of different categories of products using ARIMA, SARIMAX and FBProphet models in Python. Designed an interactive Tableau dashboard to forecast MoM, QoQ and YoY demand using above results. Created Tableau dashboard for ML-based opinion mining tool (VOX) having a detailed sentiment analysis of text reviews to facilitate the analysis of performance of the VOX engine.

Developed data transformation scripts using python and transformation libraries like Pandas, NumPy, etc. and automated them to be executed at scheduled intervals that were used to provide smooth data transfer pipeline infrastructure. Learnt to build analytics tools that utilize a data pipeline to provide actionable insights into operational efficiency and other key business performance metrics.

Learnt and worked on different containerization platforms like Kubernetes, Docker and understood the Linux boot system. Studied the effect of error factor as a feedback-type feature input to the forecasting models on data using SARIMA.