Years
Experience
Hi there! My name is Vinay Kumar, and I'm a Senior Software Engineer & Engineering Manager based in Haryana, India.
I have 10+ years of experience architecting highly scalable systems, optimizing cloud infrastructure, and leading cross-functional teams.
Proven expertise in end-to-end automation, distributed architecture, and AI-driven applications. Adept at driving technical strategy to accelerate deployment frequencies, significantly reduce operational costs, and deliver high-performance web platforms.
Deep expertise in AI agent design, RAG pipelines, and LLM integrations. Founder with hands-on experience across the full product lifecycle.

As Engineering Manager at Token Metrics, I redesigned the core cryptocurrency analytics system architecture, reducing average page load time from 7.0 seconds to 0.5 seconds and drastically improving platform availability. I led a cross-functional team of 15 engineers across Full-Stack, Data, DevOps, and QA, consistently executing sprint goals and increasing deployment frequency by 40%.
I architect and develop MeraSkool, an AI-enabled School Management (ERP) platform targeting tier-2 cities. Integrating generative AI capabilities using Vercel AI SDK, Mastra AI, and GEMINI SDK to automate support features and streamline operations. The platform scaled rapidly, successfully onboarding 26 educational institutions within the first 6 months of product launch. Additionally, I built a WhatsApp bot in just one week that eliminates the need for a parent app, allowing parents to receive notifications, ask about fees, pay online, and access results through WhatsApp.
I spearheaded the emergency rescue and delivery of the Astrobot Society NFT minting application, shipping the critical token-gating authentication feature in under one week after an outsourced team exited the project. This enabled secure analytics platform access for NFT holders, allowing them to log in to the Token Metrics analytics platform using their NFTs.
I created Add2Watchlist.com from scratch within just two weeks. This platform, built with NextJS and Tailwind CSS, offers movie recommendations based on user preferences. Leveraging OpenAI's ChatGPT, the website provides intelligent and engaging interactions to enhance user experience. As the sole developer on this project, I handled everything from initial concept and design to implementation and deployment.
I scaled the Contaqt.nl email marketing platform to process and send millions of emails daily by introducing AMQP and RabbitMQ into the data pipeline to eliminate system bottlenecks. Additionally, I developed the email flow system to track email opens, link clicks, and perform actions based on user interactions, significantly contributing to version 2 of the platform.
I architected highly available cloud infrastructure for AstraDAO using Docker and Kubernetes, leveraging auto-scaling containers to handle high-volume transaction spikes with minimal latency. This ensured the platform's stability and scalability as traffic increased, maintaining uptime and reducing latency.
I architected a robust end-to-end test automation framework using Cypress and Playwright for Change Healthcare's medical imaging web application. This reduced the manual regression testing cycle from 1 month to 1 week by fully automating browser and device cross-compatibility testing workflows. I also developed and executed comprehensive performance and API testing strategies using JMeter and TestCafe to ensure backend system reliability and latency compliance.
At Cointribe, a fintech company developing bank statement parsing for financial institutions, the team struggled with accuracy. I developed an AI model using NodeJS and Google Vision capable of handling various bank statement formats, extracting relevant data from tables within documents. The model was built and optimized within three months, significantly improving platform performance.
I was responsible for the entire development process, from design to implementation. I designed the outer plastic enclosure using Fusion 360, ensuring it was compact and aesthetically pleasing. I also developed the circuit, selecting the ESP32 board for its reliable WiFi capabilities and low power consumption. I wrote the Arduino code to manage the switch operations, and designed the web application using React and Next.js, allowing users to control the switch from their mobile devices. This comprehensive approach ensured that the smart switch was both functional and user-friendly.
At HandySolver, I reverse-engineered an obsolete medical device protocol using Burp Suite to design a custom data-retrieval server, successfully restoring access to critical stored medical data without existing documentation. The manufacturing company was no longer in business, leaving no documentation available. By intercepting and analyzing data packets, I reverse-engineered the proprietary communication protocol and developed a custom server, enabling successful data retrieval.
Token Metrics is a crypto analytics platform that displays historical data of cryptocurrencies such as Bitcoin and Ethereum, similar to a trading website. The website was initially slow, taking an average of 7 seconds to load the page and data. I redesigned and optimized the system architecture, introducing caching at different levels, including server-side and client-side caching, as well as request-level caching. As a result, the page now loads in about half a second (0.5 seconds).
Token Metrics build an NFT called Astrobot Society NFT, during the launch of NFT the team handling the minting page left and they brought me in to build the whole NFT minting page from scratch withing few days. I build the whole minting page with all the functionalities and security measures in record time.
I developed a robust automation testing framework using Cypress, JMeter, etc. The framework was able to test the application across different browsers and devices, ensuring that the application was bug-free and user-friendly.
I took on an ETL representation project that was initially very slow and unresponsive due to the large volume of data being handled by the browser, which couldn't manage it effectively. I optimized the UI to load quickly by implementing a lazy loading mechanism for the remaining data. To achieve this, I developed a custom framework, significantly improving the performance and responsiveness of the application.
During the initial phase of my career, when I had only one year of experience, I was assigned a project to retrieve data from a medical device. The device was very old, and the company that manufactured it was no longer in business, leaving no documentation available. I utilized Burp Suite to analyze the data being sent by the device and created a server based on the obsolete protocol it used. As a result, the device successfully connected to the server, allowing us to access the data stored within it.
Cointribe was developing a fintech product designed to parse customer bank statements and generate insights for banks or financial institutions. The project was initially handled by a different team that was unable to achieve the desired results. When I joined the project, I developed an AI model capable of parsing any type of bank statement within three months.
WIMWIsure, an insurtech startup, had big data and heavy traffic, but their infrastructure was struggling to handle it, causing cloud costs to rise daily. I optimized cloud infrastructure configurations and resource allocation, achieving a 50% reduction in cloud operating costs. I introduced Docker and Kubernetes, improved auto scalability, defragmented the database, and performed optimizations to reduce overall costs.
While I contributed significantly to contaqt.nl version 2, my biggest achievement was enabling the platform to send millions of emails per day. Contaqt.nl was struggling with this issue, so I introduced AMQP and RabbitMQ to solve the problem. This allowed them to scale from thousands to millions of emails per day.
They wanted to upgrade their online classes product, which previously used Google Meet. I was brought on to create a group video call service that could scale with the increasing number of students per class while maintaining video and audio quality. The product also included virtual reality features like a built-in smart class. I successfully built the product within six months.
MeraSkool.com needed a solution that would eliminate the need for parents to download any app. They wanted to remove the parent app altogether. I introduced a WhatsApp bot and built it in just one week. Now, parents receive student notifications on WhatsApp and can use it to ask for fees, pay fees online, get results, and more.
As part of my role at Skugal, I led the development of the online video call system. My work focused on enhancing the audio and video quality to ensure clear and uninterrupted communication. I also integrated AR and VR capabilities, enabling teachers to utilize 3D models during their lessons, which significantly improved the interactivity and engagement of virtual classes. By implementing these features, I helped create a more dynamic and effective online learning environment.
I optimized cloud infrastructure configurations and resource allocation, achieving a 50% reduction in cloud operating costs for this InsurTech client. I introduced Docker and Kubernetes to enhance auto-scaling capabilities, defragmented and optimized the database, which significantly improved data retrieval speeds. These optimizations ensured the platform could handle increased load without compromising performance while reducing overall costs
Maharishi Dayanand University (MDU), Rohtak | July 2012 β 2016
Active on HackerOne, Bugcrowd
"Vinay can deliver faster no matter the complexity of task" - one of my CTO.