Jan 2019 – Jan 2021 · 2 yrs · Bengaluru, Karnataka, India

• Pioneered vision of Multimodal Large Language Model & crafted 4-quarter roadmap with 25% faster time-to-delivery.
• Owned 9 Document Understanding models leading a 16-member team to deliver customized solution for 42 clients.
• Spearheaded the operational efficiency drive 2023-24 by designing org level Knowledge Graphs to automate KTLO oncall
• maintenance work, reduce manual efforts & conserve bandwidth of 50+ support engineers & in mission-critical projects.
• Contributed to leading project milestones & critical phases of ML life-cycle development across 4 Cerner Health verticals
• (Relation, Extraction, Assertion, NER) impacting 27k customers globally & enhancing user retention rate by 18%.

Aug 2018 – Jan 2019 · 5 mos · India

• Built an Object Detection Model using YOLOv3 on PyTorch interface to detect and localize antenna pole structure from video feed data (streamed from in-house cameras in Rakuten 5G Towers) and enhanced business KPIs.
• Implemented a robust scientific model to measure the bending angle of antenna pole using morphological OpenCV techniques like Canny Edge Detection & Probabilistic Hough Transform, Grab-Cut algorithm and DBSCAN based pixel clustering.
• Performed several POCs using several SOTA models & architectures on different versions of noisy data
• Obtained a model confidence score of 0.845 along with an IOU (Intersection over Union) ratio of 0.72 for the predicted class probabilities of bounding boxes after optimising model using data augmentation snd hyper-parameters.
• Deployed the model to production using TensorFlow Extended (TFX) in GCP environment.

Jan 2017 – Jul 2018 · 1 yr 6 mos · India

• Built an online, lightweight & hybrid Machine Learning model (combination of Random Forest, Gaussian Naive Bayes, Gradient Boosting) from scratch to solve Anomaly Detection in Network patterns and achieved an accuracy of 98.4% along with an RMSE of 0.33.
• Developed Deep Reinforcement Learning Algorithm for device health monitoring and fault tolerance in Cisco's Smart Router device with a confidence score of 87.86%.
• Leveraged Graph Neural Network (GNN) algorithms to generate novel Network patterns with optimized bottleneck bandwidth.
• Generated a Hand Gesture recognition model (using ResNet50 as backbone architecture) in real-time Webex videos for assisting specially abled persons; and attained an inference accuracy of 99.97% in detection of palm & hand landmarks.

Jan 2016 – Dec 2016 · 11 mos

• Completed my Bachelor’s Thesis Dissertation ”Brain Computer Interface Controlled NAO Humanoid”, an outsourced research project worth of 15,000$, in collaboration with Microsoft Research India and BITS Pilani Computer Science Dept.
• Built a functional prototype for synchronous interaction of neural impulses with a cognitive gesture-controlled robot.
• Received BITS Merit Scholarship during my thesis for perfect 10 CGPA & as a mark of Accomplishment & Innovation.
• Designed a hybrid Machine Learning algorithm using Common Spatial Pattern (CSP) algorithm, GNNs, & PCA based dimensionality reduction to extract a target stimulus from EEG; and performed Integration with real-time framework Choregraphe with 93% efficiency.

May 2015 – Jul 2015 · 2 mos

• Artificial Intelligence based Imbalance detection [for credit card fault analysis] using a hybridisation of SVM with Gaussian Kernel and other unsupervised techniques of dimensionality reduction.
• Optimised the model performance on novel data using LDA & Fisher discriminant analysis.
• Enhanced accuracy of model to 92% by using non-linear gaussian kernels and some other techniques like MCM [based on few research papers].

May 2014 – Jul 2014 · 2 mos

• Worked on an intelligent method of implementing PWM in wind and solar converter for
• MPPT algorithm.
• Used PSO, Genetic algorithm on top of a real-time simulation using Python scripts on FPGA board.
• Solved problem of overshooting minima in Gradient Descent algorithm using a self-learning adaptive learning-rate reducing oscillation frequency by 29%.

Jan 2014 – Dec 2015 · 1 yr 11 mos · India

• Teaching Assistant Executive in courses of CS Dept-
• Machine Learning
• Deep Learning
• Computer Vision
• Information Retrieval
• Neural Networks & Fuzzy Logic
• Advanced Data Mining
• Software Testing
• Computer Programming
• Microprocessor Interfacing
• Object-Oriented Analysis and Design.