profile-pic

Saravanakeerthana Perumal

I've always had an interest to learn and understand more about the world around me, which is what initially drew me to physics. My knowledge and interest gained me one of the few spots in the the Research Science Initiative program at IIT Madras in 2017, where I worked alongside brilliant scientists on a research project. Thereafter, I pursued my bachelor's degree in physics, where I grew used to looking at the world through the lenses of probability and statistics.


In my final year, I worked on a research project studying the impact of reverse migration on COVID-19 dynamics in collaboration with IIT Mandi. Applying a predictive mathematical model in python to real world data and processes was incredibly exciting to me. I gained skills in data analysis, plotting and visualization, parameter estimation and optimization, and coding in python. My background in physics gave me a strong foundation in mathematics which allowed me to understand and analyze data on a more deeper and meaningful level. I became fascinated with learning how to use software packages and techniques to unravel meaning and hidden patterns in data and how they connect with real world situations, whether it be sales data of a business or weather data from satellites.


My ability to see the world in terms of probability distributions as well as through an algorithmic perspective helped me realize data science in the right career choice for me.

  • Role

    Climate Risk Analyst & Pytorch Developer

  • Years of Experience

    2.10 years

  • Professional Portfolio

    View here

Skillsets

  • SQL - 1.0 Years
  • MySQL - 1.0 Years
  • Python - 2.0 Years
  • Git - 2.0 Years

Professional Summary

2.10Years
  • Jun, 2023 - Present2 yr 9 months

    Climate risk analyst

    Barclays
  • May, 2022 - Aug, 2022 3 months

    Research Intern

    IBM Research

Applications & Tools Known

  • icon-tool

    Git

  • icon-tool

    Docker

  • icon-tool

    AWS EC2

  • icon-tool

    Linux

  • icon-tool

    VS Code

  • icon-tool

    Flask

  • icon-tool

    MLflow

Work History

2.10Years

Climate risk analyst

Barclays
Jun, 2023 - Present2 yr 9 months
    Implemented advanced mathematical models and conducted thorough research on emerging ideas and technologies within the climate risk field, leading to the identification of innovative strategies that could potentially reduce emissions by 14% across Barclays business value chain. Engineered a proof of concept using advanced machine learning computer vision techniques to detect methane emissions in satellite imagery, enhancing data accuracy by upto 2.79 MegaTons of CO2 measurements. Conducted in-depth statistical analyses on over 500 macroeconomic variables within the scenario expansion model, identifying key trends and correlations to inform financial projections and launched an interactive dashboard using R to visualize and present projections. Increased productivity by automating time-consuming statistical tests within a sophisticated scenario expansion model using Python programming techniques.

Research Intern

IBM Research
May, 2022 - Aug, 2022 3 months
    Developed a machine learning model that accurately predicted greenhouse gas emissions from crops, enhancing the precision of agricultural emissions forecasting. Achieved RMSE of 0.08 and reduced runtime from physical model by 75%. Cleaned and optimized a large, unstructured dataset, streamlining data processing workflows and ensuring data integrity for effective model training and analysis, significantly improving data quality and reducing processing time. Collaborated closely with senior mentors and led weekly progress presentations for the project manager, delivering actionable insights and aligning on key milestones to ensure the project remained on track. Streamlined project management by organizing and documenting code in a structured GitHub repository.

Major Projects

3Projects

Effect of pretrained models on land use classification

Sep, 2024 - Present1 yr 6 months
    Used experimented tracking to test performance of pretrained ResNet50 models on the EuroSAT dataset. Implemented modularized code on GitHub and reported results.

Data Study Group Hackathon - Alan Turing institute

Feb, 2023 - Apr, 2023 2 months
    Created an object recognition model to detect litter in images in association with the charities Keep Wales Tidy and Keep Scotland Beautiful. Collaborated with 6 other researchers globally to build object detection models.

Recommendation System using Ontology/knowledge graph

Aug, 2022 - Nov, 2022 3 months
    Built a recommendation system using ontology in python. Achieved an accuracy of 87%. Presented and published paper in ETTIS international conference 2022.

Education

  • Masters in Data Science

    VIT University (2023)
  • Bachelors in Physics

    Shiv Nadar University (2021)