Venkatesh S

About Venkatesh S

Data Scientist with experience in data science, Generative AI, AI/ML, LLMs, RAG…

Experience

1. Machine Learning Engineer

   BlackRock

   Nov 2024 — Present · NJ, US

   Initiated and deployed a retrieval-augmented generation (RAG) pipeline integrating LLaMA-based LLMs with internal knowledge bases using LangChain and LlamaIndex, reducing time-to-answer for portfolio manager queries by 35% and improving decision-making efficiency. Fine-tuned pre-trained Hugging Face transformer models (BERT and GPT-4) for financial sentiment analysis on market news and earnings calls, increasing sentiment classification accuracy from 82% to 91% on validation data. Built and monitored ML workflows using MLflow and implemented CI/CD pipelines for model retraining and deployment, reducing model drift incidents by 25% and improving model refresh turnaround to under one week. Designed an LSTM-based time series prediction model on Vertex AI (GCP) to forecast daily fund inflows and outflows, improving forecast accuracy by 18% compared to legacy ARIMA models. Conducted hypothesis testing and cohort analysis on investor behavior datasets to uncover actionable insights, leading to a 12% increase in targeted client engagement for underperforming segments. Collaborated with cross-functional teams to implement model monitoring dashboards in Tableau, enabling real-time tracking of model KPIs like accuracy, drift, and response latency. Contributed to the development of a prototype generative adversarial network (GAN) to synthesize realistic synthetic market data for stress testing risk models, cutting data acquisition costs by 20%.

Education

• Saint Peter's University

  Master's degree, Data Science

• Geethanjali College of Engineering and Technology

  Bachelor of Technology - BTech, Mechanical Engineering