Built an end-to-end discrete-time simulator of an N-path receiver, with amplifier saturation, and quantization.

Derived closed-form I/O responses for signal and interference, and benchmarked against a conventional receiver.

Designed a learned MLVAMP algorithm with neural denoisers for nonlinear signal recovery under interference.

Modeled receiver impairments and built a hybrid inference framework achieving 20 dB gain over linear methods.

Developed an interpolation-based ray tracing to optimize channel modeling in Sionna.

Implemented a reflection model for accurate LOS MIMO simulation, improving spherical wavefront modeling.

Improved a SSR algorithm, and integrated into a larger coding scheme (FASURA) for massive random access.

Implemented SBL, MSBL, AMP-MSBL \& GGAMP-MSBL algorithms from scratch.

Built a neural network with Adam optimizer and early stopping, achieving 99% accuracy.

Enhanced performance through feature extraction with Librosa and hyperparameter tuning.

Developed a custom KNN distance metric, improving voter classification accuracy with incomplete data.

Enhanced model performance through feature selection, achieving over 84% test accuracy.

Analyzed a continuous-time MDP, integrating controlled actions and optimizing discrete-time observations.

Extended the framework to a multi-agent system, exploring strategic considerations in a two-player context.

Simulated a Viterbi algorithm sequence estimator, assuming knowledge of channel parameters.

Evaluated the SER of transmitted symbols for a wide range of SNR values and decoding delays.

Simulated a two-queue system with customer arrivals following a Poisson process.

Evaluated the error (< 3%) and generated time plots of the number of customers in the queues.

Evaluated the performance of naive Bayes, logistic regression, SVM & random forest classifiers.

Performed text preprocessing & word embedding (TF-IDF) on the training dataset.