Full Stack Developer

During my time at Genesis Artificial Intelligence Pvt. Ltd., I worked as a Full Stack Developer in a small but fast moving engineering team. The environment offered opportunities to learn quickly, contribute to real products, and take ownership of entire systems from the ground up. I worked on multiple AI focused and data heavy projects, and gradually became someone others relied on for structure, reliability, and delivery.

One of the most memorable projects was the speech to text data pipeline. The company was developing its own STT model and required a large volume of high quality labeled audio data. I designed and implemented a distributed scraping and processing system that collected publicly available earnings call recordings, paired them with transcripts, cleaned and segmented them, and uploaded them to an internal S3 dataset used by the AI research team. The architecture was built around RabbitMQ and Node.js, where each additional worker machine increased throughput. At peak, the system ran across eight to ten laptops at night and processed large volumes of audio almost continuously. After the initial setup, the system needed only occasional attention and dramatically sped up dataset preparation for model training.

Another major project was a support chatbot developed for EDB Bank. Dialogflow handled the AI conversations while I built both the backend and the frontend around it. The backend communicated with the bank’s SOAP based APIs to retrieve account and loan information and other service related data requested during a support conversation. Security restrictions meant the complete application had to run inside an air gapped system. To make it compatible, I created a library called mongofire which acted as a drop in replacement for Firebase but used MongoDB internally. This allowed the entire system to move from Firebase to MongoDB without rewriting existing logic. The result was a fully working end to end demo that became the base for production deployment after I left the organization.

I also worked on smaller but meaningful initiatives. For an industrial water analytics product, I built Python APIs that exposed sensor readings such as ppm, temperature and flow rate for chemical companies monitoring water usage. I later helped the AI team improve their object detection dataset generation workflow by setting up Prodigy for manual annotation tasks.

As the projects grew, I was assigned the responsibility of mentoring two junior developers. I guided them on structuring their work, maintaining consistent API standards, using version control correctly and contributing through pull requests. This improved quality across the team and reduced the burden on senior engineers, ultimately increasing delivery speed across multiple projects.

Looking back, Genesis AI was a period of hands on learning and deep ownership. I got the chance to build systems end to end, solve practical engineering constraints with creative approaches, and grow into a leadership mindset while still writing code every day.