Projects

Enigma

A tensor framework for building dynamic neural networks with optimized tensor computations and GPU acceleration.

NumPy-QuadDTye

A true cross-platform Quadruple-Precision (128 bit) floating data type for NumPy

Recursion Trace

A package to trace recursive function calls and generate/visualize the recursion tree.

Laplace Maze Solver

An innovative approach to solving mazes using Laplace's equation. Using physical principles to model the maze as a potential field and then navigate the maze by following the gradient

Gravitational Wave Sonification

Sonification of gravitational wave data

Virtual Clothing Assistant

An intelligent AI assistant allows user to try any piece of cloth virtually

MIRA - Multimodal Image Reconstruction with Attention

MIRA is a multimodal transformer (Encoder-Decoder) based architecture for Text or Image to 3D reconstruction focussing on generating the 3D representation just using single 2D image of object within seconds.

3D Cervical Spine Segmentation and Multi-Vertebrae Fracture Detection

Cervical spine 3D segmentation and fracture detection using Machine Learning to detect fractures and assess vertebrae health.

Image Style Transfer

Impose the texture/style of one image on the other without disturbing the content of the other image.

TuneX

A powerful CLI toolkit offering one shot solution for LLM in running, finetuning and instruction tuning with powerful customized kernels

PythonCoder

Code-gen model for Python based on GPT2 with MultiQuery and FlashAttention

S.E.A.R.C.H (Systematic Engine for Analyzed Retrieval and Contextual Handling)

S.E.A.R.C.H (Systematic Engine for Analyzed Retrieval and Contextual Handling) An advanced, multi-faceted information retrieval system.

Multilingual Name Entity Recognition

XLM-RoBERTa, fine-tuned in German, French, and Italian, offers robust Zero-Shot Classification with an f1-score of 0.8638 on the validation set. Its performance is enhanced by a custom Error Analysis pipeline, ensuring reliable multi-language classification.

Custom Reinforcement Learning environment for Edwin Lander

custom Gym environment built to train RL models on the Edwin Lander game to land via following the rules