Apr 2025 – Apr 2025 · 0 mo · United States · Remote

• Contributed to Euler Dataset

Apr 2025 – Apr 2025 · 0 mo · United States · Remote

• OCR ,SFT, OCR

Feb 2025 – Mar 2025 · 1 mo · United States · Remote

• Scicode, RLHF, HLE

Jul 2024 – Aug 2024 · 1 mo · Remote

May 2024 – Jul 2024 · 2 mos · Bengaluru, Karnataka, India · On-site

Feb 2024 – May 2024 · 3 mos · Remote

Aug 2023 – Dec 2023 · 4 mos · Scotland, United Kingdom · Hybrid

• Under the guidance of Prof. Sujoy Bhattacharya, my Bachelor’s thesis focused on improving high-dimensional covariance estimation by addressing the outlier sensitivity of traditional methods like Graphical Lasso (Glasso). I developed a Robust Graphical Lasso framework using the Minimum Covariance Determinant (MCD), achieving stable and sparse precision matrices even in contaminated datasets. Tested on synthetic and real financial data, the method led to a 22% reduction in estimation error and an 18% improvement in Sharpe ratios, especially in settings with over 10% noise.
• I also explored Generative Adversarial Networks (GANs) to capture nonlinear dependencies in asset returns, though Robust Glasso consistently outperformed GANs in noisy environments. This work highlights the power of combining robust statistics with machine learning for enhanced portfolio optimization and financial risk modeling.

Mar 2023 – Apr 2024 · 1 yr 1 mo · IIT Kharagpur

Mar 2023 – May 2023 · 2 mos

Jan 2023 – Apr 2023 · 3 mos

• Worked on a Portfolio optimization project. The main objective of portfolio optimization is to allocate weights to the underlying portfolio constituents to achieve a specific objective function under a given set of constraints. I have developed an algorithm that combines machine learning algorithms like particle swarm optimization and simulated annealing for solving optimization problems which give much better and more precise results than other methods like monte carlo stimulation or Lagrange multiplier.
• Some list of the Objectives functions are:-
• I. Maximize Sharpe Ratio
• II. Maximize Treynor Ratio
• III. Maximize Roy’s Safety-First Ratio
• IV. Maximize Sortino Ratio
• V. Minimize Beta
• VI. Minimize Downside Deviation

Apr 2022 – Jun 2022 · 2 mos · On-site

• Our team of five students successfully constructed an autonomous cleaning bot leveraging computer vision technology. This innovative bot operates independently without human intervention in the solar panel cleaning process. Under the guidance of Prof. Vikrant Racherla, we utilized edge detection computer vision techniques to detect and navigate the edges of the solar panel, ensuring the bot's stability and preventing potential falls

Aug 2021 – Aug 2022 · 1 yr

• Responsible for handling all work related to the Communique` website mainly focusing on the web development part.
• Leading a team of 30 students to manage a budget of 50K and conducted various career-aiding workshops and seminars for 12000+ students.
• Organized panel discussions,competitions and various internal events to improve associates' soft skills and personalities