From scratch Pytorch implementation of the original transformer paper. Briefly explain how it works through a sentence-to-sentence translation task example with English to Spanish.

Comprehensive and educational analysis of investment strategies using the performance metrics of GAFAM stocks 📈 - Google, Apple, Facebook (now Meta), Amazon, and Microsoft.
It evaluates two distinct approaches. An active strategy utilizing Moving Average Convergence Divergence (MACD) for trading signals and a passive strategy employing dollar-cost averaging (DCA) with the SPDR S&P 500 ETF Trust (SPY) as a benchmark.

Restore images with centrally masked areas using generative models to accurately predict and reconstruct missing parts.
Post-GAN training, the discriminator and generator collaboratively produce high-quality image samples using continuous gradient-based updates to the activation maps until the samples are classified as real by the discriminator.

Developed a Python-based reinforcement learning agent capable of mastering the Lunar Lander game from OpenAI Gym.
I implemented the REINFORCE loss function using categorical cross-entropy weighted by the discounted reward for each observation and balanced the need for exploration (to learn diverse strategies) with exploitation (maximizing reward outcomes).

Data-driven analysis and machine learning to detect customers who are more likely to churn. Used those insights to propose different strategies for the company to investigate for a potential retention program.

This analysis compares the implementation of linear regression models from scratch (utilizing Numpy and Pandas) with those from scikit-learn to predict the optimal selling price for a client’s house.