Shankar Kumar S

Experience

• 

  Qatar Computing Research Institute

  May 2020 - Feb 2021

  • Worked on a research project titled “Counting Cars from Space to Monitor Internal Displacement” under Dr.Ingmar Weber and Dr.Ferda Ofli.• Identified and performed a detail study on the various biases that affect satellite car detection models.• Identified and performed a detailed study on 5+ bias factors that affect satellite car detection models.• Formulated solutions to minimize bias and proved performance improvement by studying precision-recall characteristics.• Submitted a detailed report summarizing the various possible biases that a satellite car detection model can have supported by statistics to prove the bias and provided solutions for the same.• Presented at the 18th IMISCOE Annual Conference held in Luxembourg and at IC2S2 2021 held in Zurich. Show less • Worked on a research project titled “Multi task learning framework for Road Attribute Prediction using Graph Neural Networks”. • Implemented a decision tree algorithm to predict Speed Limit, One-way/Two-way, Road Heading and Average Speed attributes of a road from vehicle GPS trajectories.• Improved the algorithm to predict Oneway/Twoway road attribute without map matching using radius search.• Extensively worked in the areas of Feature Engineering. Extracted features such as distance covered, velocity, direction of heading at every GPS point.• Performed feature extraction and data pre-processing on large GPS data to enable existing Deep Learning models to work on this data. Show less

  • Machine Learning Research Fellow

    Jul 2020 - Feb 2021

  • Summer Research Intern

    May 2020 - Jul 2020
• 

  Wells Fargo

  May 2021 - Jul 2021

  Software Development Intern

  • Engineered a ML model leveraging Local Outlier Factor Algorithm and Nearest Neighbors for credit card fraud classification.• Developed Apache Kafka data pipelines to retrieve and publish streaming data and results to the model.• Utilized PySpark to develop end-to-end ETL pipeline, orchestrating data flow for model inference.

• 

  Wells Fargo

  Jul 2022 - Jul 2023

  Software Development Engineer

  • Developed a Springboot microservice to automate parallel Spark job executions, resulting in 80% reduction in manual workload.• Integrated Angular components into the financial data dashboard for visualization and reporting of job statistics to stakeholders.• Built ASP.NET APIs and SQL queries for data access and analysis features within the enterprise financial data dashboard.

• 

  University of Illinois Urbana-Champaign

  Aug 2023 - Dec 2023

  - Mentored two teams in their semester-long software project- Handled and coordinated weekly meetings and guided project development- Ensured timely completion of project requirements - Worked under Professor Yun Huang at the SALT LAB at UIUC.- Contributed to the development of an LLM research question generator chrome extension.

  • Course Assistant

    Aug 2023 - Dec 2023

  • Graduate Assistant

    Aug 2023 - Dec 2023
• 

  University of Illinois Urbana-Champaign

  Feb 2024 - now

  Graduate Research Assistant

  • Leading the development of a misinformation detection browser plugin.• Fine-tuned BERT large language model classifier in PyTorch achieving 94% accuracy for misinformation detection.• Developed Flask model inference service and React frontend to run ML model as a chrome extension.• Containerized model inference service using Docker with load balancing capabilities for enhanced scalability and achieving 74% increase in inference speed.