- Led and contributed to predicting maintenance activities for the Amazon fleet with the goal of reducingcost and operational downtime. The uptime is improved by at least 2% and cost is reduced by 16%.- Validated the fidelity of sensor data, identified the gaps in data quality, and led the efforts in collaborations with the vendor to improve the accuracy and consistency of the data within acceptablemargins.- Designed and developed a statistical model to detect tire defects using the longitudinal TirePressure Monitoring Sensor (TPMS) data and quantify tire health using a severity score that canbe used by the operation team to prioritize tire maintenance activities.- Contributed in developing a data pipeline using the DataNet Framework to consume sensor datain near real time and apply the model to detect tire health.- Led and contributed the redesign and development of a trailer recommendation engine to optimizelogistic operations and scheduled maintenance activities. The performance of the updated product isimproved by 23%.- Conducted a field experiment, evaluated the performance of the existing trailer recommendationengine, and identified the product gaps.- Proposed technology and process fixes in the product design, and contributed in deploying theupdated product in the field.

- Led and contributed to the carrier load balancing project to effectively distribute subscribers on satellite carriers with the goal of improving service quality.- Developed an analytical model to estimate the probability of speed availability for a scheduler supporting weights and priorities using the Queuing theory.- Designed and developed a greedy algorithm to dynamically distribute subscribers over carriers with the goal of improving the quality of delivered services.- Contributed to developing an algorithm to passively estimate offered speed of an Internet connection using the Inference theory.- Led and contributed to the causal and experimental analysis of the service feature testing.- Designed an interleaved cross-sectional and longitudinal experiment to evaluate the effectiveness of the load balancer in the field with the goal of minimizing seasonal effects.- Performed power analysis to identify the number of samples required to gain statistical confidence in the experimental outcomes.- Performed hypothesis testings on the outcomes of traffic engineering features in the field using statistical methods.- Contributed in designing differential traffic scheduling policies for a weighted fair scheduler with the goal of improving speed availability and minimization of SLA violations.- Led and contributed to the video quality of experience estimation project for the encrypted network traffic.- Contributed in developing a Chrome browser Java Script plugin to collect video quality of experience metrics for YouTube and NetFlix content providers.- Performed feature engineering and identification using a deep neural network-based video QoE estimation model. - Designed a stochastic model; a variant of SARIMA to predict the timeseries of service usage.- Designed a discrete choice model incorporating latent variables for predicting subscriber demands using the Microeconomics theory.

- Proposed a novel concept of building an optimization framework using which congurable optimization applications can be built to provide optimization as a service.- Led, and contributed to the optimization framework project, delivered the project in production as a part of the AT&T ECOMP platform, and contributed in open sourcing the project in ONAP.- Took the project from soup to nuts; designing of the functional and software architectures of the framework, managing the agile development process as a scrum master, developing the library of software functionalities, documenting the interfaces and deployment/operational guidelines, assisting the testing team by providing test-cases and narrowing down the root causes of failures, and assisted the production teams in deploying the framework in production.- Convinced the ONAP community on why the optimization framework should be a part of theONAP ecosystem, open sourced the project, and contributed in enhancing the framework functionalities.- Led the project on identifying the cutting-edge technologies to specify, evaluate, and enforce stateful policies in the ECOMP/ONAP policy platform.- Identified Drools as a technology of choice that is employed as a core module of the policy platform, where my specific contributions were; analyzing underlying algorithms, operations, and software architecture of Drools, evaluating the capabilities and limitations of Drool's operational modes, and conducting stress tests and experimentally analyzing the performance of the engine.- Modeled a rule condition and designed an upper bound on potential matchings, which is used to predict and monitor the memory consumption of Drools.- Designed a solution to achieve scalability and resiliency in the policy platform using the Kubernetes platform in ONAP.

- Designed and developed a set of tools using innovative optimization algorithms and visualization techniques. The tools are actively used by AT&T planners to design multi-layer (IP/OTN/WDM) networks, home a customer, find a route for a service, and analyze vulnerability.- Applied unsupervised learning techniques to design an algorithm for the placement of hardware and software capabilities in a data center to maximize sharing of cloud resources while optimizing end-to-end service performance.- Proposed and developed algorithms to identify diversity violations in metro networks to perform network impact analysis, and designed novel graph theory-based tunnel reconfiguration algorithms for congestion avoidance and mitigation during maintenance windows in a backbone network.- Introduced unique visualization capabilities to correlate a multi-layer network topology, traffic distributions of services, and geography data to analyze a data center and network placement solutions, and designed user interfaces to perform vulnerability analysis and customer homing.- Proposed and designed a novel L2-VPN protocol to minimize the required memory in switches to support Layer-2 VPN, built a testbed using the mininet emulator, and demonstrated the feasibility of L2VPN service using the proposed protocol.- Designed and developed a controller (using RYU SDN framework) with network discovery, traffic monitoring, and VLAN/MPLS tunnel provisioning functionalities to program open-vSwitches and provision a service.- Demonstrated closed-loop operations to support the network growth use case in which my specific contributions were, developing an application over the ODL-SDN controller to orchestrate the capacity augmentation/subtraction process in multi-layer (IP/OTN/WDM) networks, building traffic monitoring and explicit routing functionalities, and implementing a user interface for policy specification and network state visualizer.

- Designed a hybrid electro-optical data center architecture and an application to route elephant and mouse flows in such data centers. A prototype is developed using NEC-PF5240 switches,MEMSs, and the TREMA controller.- Designed an SDN controller for packet-over-optical networks. Developed an SDN-application for traffic engineering, a homing application for mapping cloud demands in data centers, a scheduling application for optical burst transmission, and a congestion-aware transmission-rate controller application for congestion-avoidance in TCP/IP networks.- Developed a burst assembling application that can be used as a network function in a commodity server to realize optical burst switching.- Designed routing and spectrum allocation applications and protocols to support time-varying traffic over flexible-grid WDM transport networks.- Designed ROADM node architectures with unicast/multicast and optical/electrical traffic grooming capabilities for flexible-grid WDM networks.- Designed optimization algorithms for network defragmentation, optical/electrical traffic grooming, dedicated/shared protection, and waveband switching in flexible-grid WDM networks. Derived bounds and performed complexity analysis to evaluate the efficiency of the proposed solutions.- Derived analytical models to estimate the blocking probability of optical connections in fixed and flexible-grid WDM networks.- Designed architectures and prototypes for Flexible-Ethernet and Flexible-OTN to enable flexibility in client interfaces of a transponder.

- Designed optimization algorithms to provision survivability at both IP and WDM layers with dedicated and shared protection, and developed an event-driven simulator to evaluate their cost-performance trade-offs.- Designed an optimal algorithm for the placement of optical-2R and OEO-3R regenerators in single line-rate networks and a heuristic algorithm for regenerator site selection in mixed line-rate WDM networks.- Designed architectures for advance reservation switching, such as time-shift and variable-bandwidth circuit switching to provision Quality of Service (QoS), and developed efficient algorithms to maximize network throughput.- Designed a supervised learning algorithm to identify anomalous network behavior using historicaldevice configurations data.- Applied the collaborative filtering machine learning algorithm to recommend movies.- Designed a deep neural network to recognize hand-written digits.- Designed an analytical model for a QoS-enabled wireless packet switching system, and evaluated its performance by developing an event-driven simulator.- Established sensor network communications between the sensors and the gateway using ZigBee, and transferred data to a remote computer using Wi-Fi (IEEE802.11) and to a mobile handset using GPRS/GSM.

- Assisted undergraduate and graduate students on the theoretical and programming assignments for EE/CS courses such as Introduction to Telecommunication Network, Optical Communication Systems, Introduction to Optical Systems, and Electromagnetic.- Designed and setup experiments for Communication Theory Lab, Electric Network Lab, Senior Design Project Labs.- Conducted problem solving sessions, mentored Students, graded quizzes, and proctored exams.

- Designed and developed a control module and its software in an assembly language to control 64-microwave phase-shifters.- Designed special-purpose microwave components using ANSOFT, such as rectangular waveguides, circular waveguides, a power absorber, and rectangular-to-circular waveguide converters.- Analyzed the performance and behavior of these components using the ANSOFT simulator and Vector Analyzer.