Jul 2021 – Jul 2024 · 3 yrs · Bengaluru, Karnataka, India

• My role encompassed leading a comprehensive Vue migration from an outdated MVC framework to a modern architecture, leveraging cutting-edge technologies like Pinia for state management and optimizing the initial load process. This transformation was aimed at enhancing performance, improving code maintainability, and facilitating a more agile development environment. I spearheaded the development of real-time features to enhance user interaction and engineered a sophisticated data response service to improve data handling efficiency. My efforts extended to refining store management practices and overseeing code quality through diligent PR reviews and bug resolution, significantly advancing the project's overall quality and efficiency.
• Development Efficiency: Reduced development cycles by 2 days per engineer, directly translating into faster project turnaround times.
• Performance Improvements: Achieved substantial enhancements in site performance metrics, including reducing the LCP from 11s to 5s and the FCP from 5s to 2s, which greatly improved UI/UX responsiveness and positively impacted site SEO and user engagement.
• User Interaction: The introduction of real-time features and efficient server load management led to more responsive and engaging user interactions.
• Data Handling: Through the use of advanced technologies for data aggregation and storage, I managed to significantly optimize API call efficiency and page responsiveness, which contributed to a more seamless user experience and supported the onboarding of over 20k new teachers.
• Productivity and Code Quality: The adoption of Pinia for store management and the focused effort on PR reviews and bug fixes resulted in a marked reduction in development time and a 50% weekly decrease in bugs, enhancing productivity, system reliability, and accelerating the release of more robust and efficient software solutions.

Jul 2019 – Jul 2021 · 2 yrs · Bengaluru Area, India

• I managed VMware vSphere environments, emphasizing detailed Root Cause Analysis for vCenter and ESXi to improve performance. My optimization of the VMware infrastructure setup, including VM/resource allocation and automated certificate management, led to enhanced cloud integration and resource efficiency. Additionally, I enhanced vCenter and ESXi security and operational efficiency through better upgrade processes and advanced certificate management scripting, resulting in strengthened system security and streamlined maintenance processes.
• Improved System Reliability: Reduced incident response times, enhancing reliability and uptime.
• Enhanced Infrastructure Efficiency: Achieved better cloud integration and resource utilization, boosting performance.
• Robust Security and Maintenance: Strengthened security and simplified processes, ensuring more efficient operations.

Oct 2018 – Jun 2019 · 8 mos · Bengaluru Area, India

• I was instrumental in developing a cutting-edge end-to-end pipeline for employee facial recognition, encompassing everything from data collection to authorization. This project involved the enhancement of the RESNET model accuracy through meticulous fine-tuning with Transfer Learning techniques. Additionally, I pioneered the creation of an application designed for comprehensive facial data capture, utilizing OpenCV and MTCNN for data gathering, and subsequently trained the model on the YOLOv3 SSD network to improve angle detection capabilities. My work focused on optimizing deep neural network architectures and applying Transfer Learning with models like Inception-Net on custom datasets, including dog breeds and facial data, to push the boundaries of machine learning applications in authentication and recognition technologies.
• Facial Recognition Accuracy: Through the fine-tuning of the RESNET model with Transfer Learning, I significantly enhanced the facial detection and recognition capabilities, achieving an 89% accuracy rate. This advancement notably refined the accuracy of employee authentication processes.
• Advanced Detection Performance: The application developed for facial data capture, when trained on the YOLOv3 SSD network, attained an 80.19% mean Average Precision (mAP), outperforming both Faster R-CNN and YOLOv2. This breakthrough led to faster and more precise user authentication, showcasing the potential of optimizing deep neural network architectures for real-world applications.

Jun 2017 – Nov 2017 · 5 mos · Kolkata Area, India

• I developed an IoT product that integrates sensors for humidity, temperature, soil moisture, pH, and nutrients, enabling precise agricultural management through wireless server connectivity. This solution provides real-time environmental monitoring, with data visualized on a website using Node-RED, InfluxDB, and Grafana for comprehensive insights. This streamlined approach enhances agricultural decision-making and resource optimization, significantly advancing crop management practices.
• Water Usage Efficiency: By enabling precise irrigation based on real-time soil moisture and weather data, water usage has been reduced by up to 30%, conserving vital resources and reducing costs associated with over-irrigation.
• Crop Yield Increase: The optimal environmental conditions maintained through the system have contributed to a 20% increase in crop yields, enhancing food production without additional land cultivation.
• Reduction in Fertilizer Use: With nutrient sensors providing detailed soil chemistry data, fertilizer application has been optimized, leading to a 25% reduction in usage. This not only lowers the cost of inputs for farmers but also minimizes environmental pollution.