About Vibhanshu
I am a full stack developer who also trains the models. On the web side that means React, Next.js and TypeScript on the front, with Node and Python behind them. On the other side it means YOLO, OpenCV, TensorFlow and PyTorch, plus the unglamorous half nobody puts in a job title: collecting the images, cleaning the labels, and running training again after the first result turns out to be worthless.
My journey
I started with web development, which is still where most of my instincts come from: if a thing cannot be opened in a browser and used by someone who did not build it, it is not finished. That is the part of the work I care about, and it is why the machine learning projects here all have an interface rather than sitting in a notebook.
The models came later and turned out to be the same discipline applied to messier material. There is no framework that will tell you your labels are inconsistent, or that the model has learnt the background instead of the object. You find that out by looking at the failures one by one. Training my own YOLO checkpoints, on a dataset I collected and cleaned rather than one that arrived pre-packaged, taught me things no syllabus covers.
I led a team at Rajasthan Police Hackathon 1.0 on Problem Statement #6: geo-tagging privately owned cameras, so investigators can work out what a street can actually see and who to ask for the footage. I was the AI developer on it as well as the lead, and we took it through to the final stage in Jaipur. It is the project that made me care about consent and access control as design problems rather than paperwork.
Since then I have kept the two halves together on purpose. I would rather be the person who can train the model and then ship the thing around it than be handed one and asked to wire it up.
What I work on
Web Development
Crafting scalable and interactive web applications.
Computer Vision
Training detection and segmentation models, and the data work behind them.
Data Automation
Streamlining workflows and extracting value from data.
Prompt Engineering
Optimizing AI interactions through strategic prompt design.
AI & Machine Learning
Building intelligent systems for automation and insights.
Technical expertise
Frontend
- JavaScript
- HTML5
- CSS3
- React
- Next.js
Backend
- Python
- Node.js
- Express.js
Database
- MongoDB
- MySQL
Cloud
- AWS
- Google Cloud
- Vercel
- Netlify
- Firebase
AI / ML
- TensorFlow
- PyTorch
Tools
- Docker
- VS Code
- GitHub
- npm
- Figma
Where I've worked
Full Stack Developer · Aptness.ai
Gurugram · Sept 2025 to now
- Building AI-driven workflow automation on a platform that lets teams automate business processes end to end
- Connecting backend APIs to React interfaces to ship scalable SaaS modules
- Cutting average load time through query optimisation and caching
- Working in Agile sprints with Git and CI/CD pipelines
Technical Intern · MRB Enterprises
Mumbai · May to July 2024
- Administered digital tools and improved day-to-day process efficiency
- Reduced manual data entry errors through better validation
- Handled troubleshooting and system updates to keep downtime low
Technical Head · Super AI Community, TCET
Mumbai · Mar 2023 to July 2024
- Led technical operations for the college's AI community and its stakeholder events
- Managed system setup, connections and data for community programmes
- Provided technical support and ran system updates across events
Education
B.Tech, Artificial Intelligence and Machine Learning
Mumbai · 2025
Thakur College of Engineering and Technology/CGPA 9.17
Let's connect
I'm always open to discussing new projects, collaborations, or just talking through an idea.