Every six months, in six countries around the world, EF brings together and funds individuals to help them meet their co-founder, develop an idea and raise money from leading investors - as fast as possible.Part of 1st Graduate Cohort• Engaged with a curated community of exceptional future founders to develop ambitious ideas into real products• Participated in multiple Generative AI hackathons & AI Safety initiatives, exploring latest advancements

Adversarial Learning for Online Malware Detection• Built an Adversarial Learning based algorithm for dynamic Malware Detection to generate a diversity of attacks constantly mis-training the classifiers by changing the data distribution & features importance.• Developed an online learning model using River package for continuous performance evaluation handling all time evolving & distribution changes in the attack stream improving accuracy for unseen attacks upto 89%.• Designed experiments to test the integration of conditional GANs for Network Traffic Anomaly Detection.

Conducted weekly tutorial sessions, solving coding problems and demonstrating model training challenges

Automation of Financial Model Risk Validation• Designed a Recommendor System for model risk related Keyword Extraction & predictive Text Analysis• Implemented Probabilistic Topic modelling using LDA to discover hidden structures in historical Risk Findings• Formulated Equity Trading & Investment Planning problems in Reinforcment Learning frameworkand delivered talks on Multi-Armed Bandits, Markov Decision process to their Academic Research group

A team of 40+ students which designs and fabricates Mars Rover Prototypes• Designed an Augmented Reality Tag detection algorithm using Iterative Search Techniques and SpatialThresholding based Crater mapping by RPLidar for the University Rover Challenge(URC), 2021• Developed a way-point follower pipeline combining a global planner working on A* algorithm with a localplanner that uses a Dynamic Window approach & future Trajectory Estimation for autonomous traversal• Devised a robust arrow detection architecture using OpenCV for contour estimation, polygon alignmentand corner refinement with Landmark 3-D mapping algorithm for rover’s localisation in real time

Road Network Extraction with Deep Layer Aggregation• Developed a road network inference architecture combining Semantic segmentation with Graph Theory• Introduced Graph-Tensor Encoding (GTE) that maps road graphs to a tensor representation andfacilitates the training of non-recurrent, supervised model to capture graph structure directly from an image• Surpassed prior methods on two widely used metrics- TOPO & APLS and demonstrated that our approachwas capable of robustly inferring stacked roads (e.g., overpasses) that lacked by previous implementations• Migrated the source Tensorflow 1.xx implementation with ResNet and U-Net backbones into PyTorch

• Engineered a hybrid architecture consisting Time-Distributed Encoder with Attention Blocks to classifyacceptance of Trade Offers and a Gradient Boosting Regressor to capture the Maximum accepted prices• Configured Forecast Horizons and target lags creating Partial Auto-correlation Function (PACF) Plots• Implemented an Online Learning based ConvLSTM Network for highly imbalanced Energy Trading Data

Developed a Deep Convolutional Network and Image Processing based Pipeline to accurately Align Farm Patches and commute points of using raw Geospatial Data received through IOT devices. Implemented Genetic ALgorithm for Optimising the model and other Statistical techniques for protrusion detection and correction and Gradient based segmentation.