Amar Babuta
Development
Delhi, India
Skills
Data Science
About
Amar Babuta's skills align with IT R&D Professionals (Information and Communication Technology). Amar also has skills associated with Programmers (Information and Communication Technology). Amar Babuta appears to be a low-to-mid level candidate, with 3 years of experience.
View more
Work Experience
Data Scientist
Veolia Australia and New Zealand
October 2021 - Present
- Melbourne Responsibilities: * Implementing pipelines to perform data cleansing, extraction, processing, storage, and manipulation from multiple data sources provided by the client stored in form of a database on the cloud and finding correlations to observe customer behaviour. * Analysing data to validate third-party solutions by comparing results from the software with the operator assessment of defects in the sewer pipes. * Created an AI model to predict the likelihood of blockages in the sewer network using a different set of features like Material, Diameter etc. * Identify failures and break testing of cloud-based third-party solutions to enrich Veolia's digital capability. * Key technologies used - Python, Google Cloud Platform (GCP), Big Query, FastAPI, Pydantic models. * Contribute to the design, build, and implement cloud-based artificial intelligence models to improve operations of treatment plants and networks. * Identify failures and break testing of cloud-based third-party solutions to enrich Veolia's digital capability. * Accountable for code production, including constant maintenance of the code (object-oriented) and re-factoring when necessary. * Produce user-friendly interfaces to share results with end clients, e.g., interactive maps or dashboards, etc. * Participate in major tenders, for example by providing detailed data analysis to show-case Veolia capacity in data analytics/AI. Achievements * GreenPath Online - Implemented the first instance of GreenPath Online, the Veolia solution automatically tracking emissions on a monthly basis. This solution significantly helped manager and operators be more aware of the carbon footprint of their activities in water treatment plants. This solution has helped operators in targeting the activities and to come up with a plan which will help them in reducing the carbon footprint in 2024 by 7%. * Virtual Submetering Trial - Successfully validated the Veolia solution Virtual Submetering which splits energy consumption by equipment in a given facility. This trial proved that the solution is applicable in an Australian context and also show that performing energy assessment by labour takes 28 days to record all values whereas using the solution reduces labour hours by 14 days. * AI model Industrialisation - Developed and implemented a range of AI solutions bringing process efficiency throughout Veolia operations. In particular, the solution BurstID was found successful to optimise pipe renewal plans to minimise water losses from water networks in Australia and New Zealand.
Data Scientist
The University of Melbourne
March 2020 - December 2020
- Melbourne Responsibilities: * Implemented Data exploration on 1.5GB of data to identify key features out of 54 features for the admission data like finding correlations between different workflow processes of the applications and the time taken for the workflow process to complete. * Performed experiments daily by performing data cleansing, processing, and implementing Machine Learning classifiers using Python to predict the likelihood of the application becoming successful after the offer is released and the amount of time it takes for the applicant to accept an offer. * Refined Hyperparameters of the models like Decision Tree, and Random Forest classifiers in Python so that the predictions made by the model are as accurate as possible. * Developed an optimal combination of workflow processes for application to improve turnaround times for The University of Melbourne to review the assessment. * Communicated the results effectively with the analytics department with the help of an interactive dashboard and visualization using Graphviz, and Shiny to enable tracking of admission process average and median times with the help of flow charts. * Key technologies used - SQL, Python, Power BI, and Tableau.
Education
The University of Melbourne
Master of Data Science
February 2019 - December 2020
