projects

Featured personal projects—AI agents for investment research and LSTM-based market modeling.

Selected work from my GitHub. For more repositories, see repositories or github.com/jbsdoki.

Jump to: ADK agent · LSTM model

AI agent for stock market investment research

Jun 2025 – Jul 2025 · Personal project

Built with the Google Agent Development Kit (ADK) and Gemini: an agent that searches the web and uses the Yahoo Finance ecosystem (via yfinance) to gather data for stock research—historical performance, forward-looking context, and current valuations.

Architecture: a manager agent coordinates flow, with three sub-agents focused on past performance, future performance outlook, and current valuations and market data.

Skills: Google Agent Development Kit (ADK), Gemini, Python, web scraping, Yahoo Finance / yfinance

jbsdoki/Google_agentic_AI_experiment

LSTM machine learning model for stock prediction

Jan 2025 – Feb 2025 · Personal project

A TensorFlow Keras LSTM model trained on historical market data: given a window of N trading days, the model predicts direction or behavior for day N + 1.

The repo covers experimentation, data preparation with Pandas, and training/evaluation workflows for time-series forecasting.

Skills: TensorFlow, Keras, LSTM, Pandas, Python

jbsdoki/ML_Market_Prediction