Ronak Sheoran

Data Scientist | AI/ML Engineer
Bangalore, IN.

About

A highly analytical and results-driven Data Scientist with a strong foundation in Machine Learning, Financial Engineering, and advanced statistical modeling. Proven ability to develop and deploy production-ready AI/ML solutions, optimize trading strategies, and extract actionable insights from large datasets to drive significant business and financial outcomes. Eager to leverage expertise in quantitative research and data science to contribute to innovative projects in dynamic environments.

Work

udaan.com
|

Analyst - Data Science Intern

Bangalore, Karnataka, India

Summary

Led data science initiatives at udaan.com, developing advanced AI/ML models and performing rigorous data analysis to optimize business processes and enhance decision-making.

Highlights

Engineered a production-ready NL2SQL model leveraging graph-based CTE retrieval, RAG, and parallel multi-run generation, optimizing for minimal token usage and robust, hallucination-resistant deployment.

Developed a Deep Q-Network to dynamically optimize inventory item discounts based on expiry, balancing maximum revenue generation with minimum wastage.

Conducted A/B and time-series analysis with paired t-tests to evaluate CPOD rollout impact on AUM, providing statistical and practical insights into business performance.

Constructed and maintained Retention Matrices using Spark SQL for datasets up to 3 billion rows, automating Supply Chain Cost calculations and visualizing results with Excel and Power BI.

Quant Insider
|

Quantitative Researcher, Part-time

Remote, Virtual, India

Summary

Developed and backtested algorithmic trading strategies, contributing to advanced study materials and improving client confidence in technical interviews.

Highlights

Developed and backtested algorithmic trading simulations and strategies, providing practical case studies for training purposes.

Researched and documented cutting-edge quant finance strategies, translating complex findings into advanced study materials.

Contributed to significant improvements in client confidence for technical interviews through specialized content.

Soul AI
|

RHLF, LLM, Data Scraping

Remote, Virtual, India

Summary

Contributed as a domain expert to state-of-the-art Large Language Model (LLM) training and development, focusing on Reinforcement Learning via Human Feedback.

Highlights

Acted as a domain expert in Reinforcement Learning from Human Feedback (RLHF), contributing to the training and fine-tuning of state-of-the-art Large Language Models (LLMs).

Collaborated with a cross-functional team of 50+ members to develop a cutting-edge AI model, leveraging expertise in Probability and Statistics.

Enhanced the overall effectiveness of AI solutions by contributing to over 500 prompts, ensuring robust model performance.

Education

Delhi Technological University
New Delhi, Delhi, India

Bachelor of Technology

Mathematics and Computing Engineering

Grade: CGPA: 8.4

Courses

Machine Learning

Probability and Statistics

Financial Engineering

Stochastic Processes

Stochastic Calculus

Regression Analysis

Linear Algebra

Data Structures and Algorithms

Operating Systems

DBMS

Languages

English

Skills

Programming Languages

C++, Python, R, SQL, MATLAB.

Frameworks

PyTorch, PySpark, Langchain, Pandas, NumPy, Scikit-learn, Statsmodels, Seaborn, Matplotlib.

Tools & Technologies

MS Excel, Power BI, Databricks, Git/GitHub.

Data Structures & Algorithms

LeetCode, GFG, Data Structures, Algorithms.

Quant Finance

Algorithmic Trading, Portfolio Optimization (Markowitz model and CAPM), Options Pricing (Binomial Model, CRR model, Black-Scholes).

Machine Learning

Neural Networks, Deep Reinforcement Learning, LSTM, CNN, XGBoost.

Projects

Pairs Trading Using Copula Mixture Model

Summary

Developed a copula-based pairs trading strategy to model non-linear dependencies between asset pairs, outperforming traditional Bollinger Bands methods on key metrics.

Time Series Forecasting Using LSTM

Summary

Developed a predictive model using Long Short-Term Memory (LSTM) neural networks to forecast future stock prices based on historical data.

Mean Reversion Pairs Trading Strategy Using Kalman Filter

Summary

Implemented a Mean Reversion Pairs Trading strategy, incorporating the Kalman Filter technique instead of a moving average approach.