Gurarmaan S. Panjeta's Projects
Implemented a deep neural network (DNN) inference for classifying across 12 audio keywords. Given the feature vector of a 1 second audio clip, the API returns the top 3 keywords with highest softmax probabilities.
A Congestion Control Protocol - TCP NewRenoPlus, and implementation and Simulation of a variable cached P2P network
An Electrocardiogram (ECG) signal shows the electrical activity of the heart in terms of quasi periodic voltage signals. The signal were studied and processed to analyze the activity of the heart. We measured the heart rate and flagged heart abnormalities or arrhythmia by ECG signal analysis.
an efficient ranked retrieval system for English corpora, optimised with VBE and BPE.
A 2-player game designed for COP290
ML algorithms from scratch
Basys3 implementation of an MLP trained on the MNIST Dataset
Exploring Imitation Learning (DAGGER), RL (Policy Gradients and Soft Actor-Critic) and Imitation-Seeded RL for training MuJoCo Environments in OpenAI's Gym
A collection of resources to get acquainted with the field of Planning and Learning in the robotics world, with strides towards incorporating LLMs and end-to-end neural encodings to make better decisions
A Descriptive Landing Page for the Github profile
Parallel Implementations of LU decpompositions, leveraging (i) pthreads and (ii) openMP
Multi-Modal Search over fashion datasets. Information Retrieval Tasks optimised using Spotify's Annoy, using text embeddings over captions and words generated from image via classifiers. Experimentation and Optimisation of hyper-parameters
State Estimation using Kalman Filters, EKF and solving the Data Association problem
Analysis and implementational details of rank aggregation methods for combing results of multiple engines to achieve best mAP and P@5,10 values
Psuedo Relevance-Feedback, Reranking using Language Modelling (RM1) and Query Expansion (word2vec, local embeddings)
xv6 schedulers and Character devices for the linux Kernel
SciPy library main repository
Analysis, Implementation and Testing of Totally Ordered Multicast built over reliable unicast, for applications that may require atomic and totally ordered multicasting facility