I primarily work with Python and Node.js, designing APIs, data pipelines, and services that are easy to reason about and hard to break. I care a lot about system clarity — good abstractions, predictable behavior, and code that the next engineer won’t hate you for.
What I Work On
Over the last few years, I’ve worked on systems where performance and correctness weren’t optional. I’ve built backend services for government platforms, AI-powered internal tools, and production-grade applications that serve real users every day. I’m especially interested in problems where you need to combine traditional backend engineering with machine learning or computer vision — not as a research exercise, but as something that has to run continuously and reliably.
A recurring theme in my work is making slow or manual systems fast and automated. Whether it’s cutting API response times from minutes to seconds, or removing human intervention entirely from evaluation workflows, I enjoy finding the bottlenecks and fixing them properly.
Selected Work
One of the more challenging projects I’ve worked on is an Automated Driving Test System for the Gujarat Motor Vehicle Department. This involved designing real-time backend software that ingests multiple RTSP camera streams, runs computer vision pipelines, and evaluates driving performance automatically. The system completely eliminated manual assessment, handled noisy real-world conditions, and delivered instant results at scale.
I’ve also built an LLM-based quality assurance system for call centers, where the goal was to automate evaluation workflows that were previously handled manually. This system reduced effort by nearly three times while improving consistency and turnaround time. While working on this, I spent a lot of time optimizing retrieval pipelines and prompt orchestration to make sure responses were both fast and reliable.
In the health-tech space, I contributed to a digital health platform under India’s National Health Mission, focusing on backend performance and data retrieval. By redesigning database queries and aggregation pipelines, we were able to reduce response times from seconds to milliseconds, which had a noticeable impact on usability for field operators.
Outside of these, I’ve also worked on e-commerce systems with integrated loan disbursement flows, combining backend services with modern frontend frameworks to deliver end-to-end solutions.
Tools & Technologies
My day-to-day work revolves around Python (FastAPI, Flask, Django) and Node.js, with PostgreSQL and MongoDB as primary data stores. On the AI side, I’ve worked with LLMs, RAG pipelines, Transformers, PyTorch, and TensorFlow, mostly in applied settings where latency and cost matter. I’m comfortable deploying systems on AWS and Azure, and I rely heavily on Docker, Git, Nginx, and Linux to keep things predictable in production.
Outside the Code
I like understanding systems deeply — not just how to use a tool, but why it behaves the way it does. That mindset has carried over into my academic work as well, including research projects in computer vision, audio classification, and my master’s thesis on Graph Language Models and structured knowledge reasoning.
When I’m not coding, I’m usually reading about system design, experimenting with side projects, or refining things I’ve already built.
Connect
If you’d like to talk about backend systems, applied AI, or interesting engineering problems, feel free to reach out.
GitHub: github.com/yashvardhan-verma
LinkedIn: linkedin.com/in/yashvardhan-60b3a5184
Email: iamyashvardhan6@gmail.com