About

Hi, I'm Subhasish, an Data Scientist based in India, with an experience of 1.5 years in the field of Data Science and Finance. I am a dedicated professional with a solid foundation in Mathematics and Data Science. My expertise spans Machine Learning, Data Analysis, Deep Learning and Mathematics. I hold a Bachelor of Science degree with a honours on Mathematics and a Masters of Science in Data Science. My passion for quantitative analysis, coupled with my commitment to harnessing data-driven insights, allows me to excel in solving complex problems.
While I have experience in the financial industry, my true passion lies in exploring new technologies and growing my skills in Data Science, regardless of the domain. In addition to my quantitative finance background, I have previously worked on tasks related to Computer Vision and Natural Language Processing, broadening my experience in the field.

Recent Work:
- Developed a high-frequency trading strategy backtesting engine to analyze and optimize trading strategies for maximizing profits.
- Developed a ML based portfolio allocation algorithm based on risk-taking capacity and financial goals.

Explore some of my notable work and projects in the links below!

Floating Image

Contact

Skills

Primary

  • Python
  • Machine Learning
  • Deep Learning
  • Reinforcement Learning
  • SQL
  • Docker
  • NLP
  • Computer Vision
  • LLM

Secondary

  • Data Mining
  • Data Visualization
  • Data Analysis
  • Time Series Analysis
  • Git/GitHub
  • Tensorflow
  • Pandas
  • Numpy
  • Scikit Learn

Soft

  • Communication Skills
  • Attention to detail
  • Adaptibility
  • Problem Solving
  • Collaborative Work
  • Time Mangement
  • Creativity
  • Presentation Skills

Experience

  • Jr. Data Scientist, Cloudcraftz Solutions

    July 2022-Present

      Options Backtesting Engine:

    • Developed a high-frequency trading strategy backtesting engine to analyze and optimize trading strategies for maximizing profits.

    • Designed and implemented the portfolio module for real-time trade storage and updates.

    • Utilized the Black-Scholes method to generate synthetic option prices for unavailable market data.

    • Employed classical machine learning techniques to enhance the strategy, including supervised and unsupervised methods.

    • Created a classification model to predict trading signals based on implied volatilities and realized volatilities.

    • Employed hierarchical and K-means clustering for trend analysis and trade quantity control.

    • Achieved a 25% improvement in PnL over the baseline strategy using supervised techniques and a 34% improvement using unsupervised methods.

      • Portfolio Allocation using Machine Learning:

    • Worked on a project for portfolio allocation based on risk-taking capacity and financial goals.

    • Leveraged Classical Machine Learning for future price prediction of stock time series data, resulting in accurate price forecasts.

    • Explored reinforcement learning and deep learning approaches for portfolio optimization.

    • Employed portfolio optimization techniques, including max sharpe and min vol methods.

  • Data Science Intern, ISI Kolkata

    May 2022 - Jul 2022

      Incidental Scene Text Detection:

    • Engineered an efficient model that excels at accurately detecting textual regions within images.

    • Employed the ICDAR 2015 and ICDAR 2013 datasets to train and validate the model, ensuring robust performance.

    • Utilized rectangular bounding boxes for precise text region localization.

    • Developed a model based on ResNet-50 architecture, demonstrating its effectiveness in efficiently detecting text regions within images.

    Data Science Intern, IIT Patna

    Dec 2021 - Apr 2022

      • Cyberbully Detection on Multi-modal Indian Languages:

      • Developed a model that predicts the sentiment and bully class of tweets/sentences in various Indian languages.

      • Collected a dataset by scraping 6436 random tweets from Twitter.

      • Compiled a Hindi-English corpus based on the collected tweets.

      • Designed and built a model utilizing BERT and BiLSTM architectures, achieving excellent accuracy in classifying sentences.

    Download Resumé

Personal Projects

  • Notebook Renderer Library

    • Developed a Python library hosted on PyPI for rendering and playing YouTube videos within Jupyter notebooks.

    • Implemented modular coding practices for maintainability and scalability.

    • Integrated GitHub testing for automated testing of releases.

    Student Performance Indicator

      • Conducted data collection, exploratory data analysis (EDA), and feature engineering.

      • Developed a predictive model for students' test scores considering variables like Gender, Ethnicity, Parental Level of Education, Lunch, and Test Preparation Course.


    Flight Price Prediction

      • Implemented a model for predicting flight fares between source and destination based on date, time of arrival and departure.

      • Applied machine learning techniques, regression, exploratory data analysis (EDA), feature engineering, and model selection.