I build the dashboards that explain the numbers, the models that predict what's next, and the interfaces people actually enjoy using — currently at Nimbus, previously shaping email campaigns at KRAFT.
I'm a data scientist and designer based in Kathmandu, currently splitting my time between machine learning models and brand visuals at Nimbus. I started in design — packaging, email campaigns, brand systems — and moved into data because I wanted to know what the numbers were actually saying before I designed around them.
A DataCamp Fellow at Code for Nepal and a certified Professional Data Scientist, I now work across the full pipeline: clean the data, train the model, ship the dashboard, and make it look like something people actually want to open.
Collaborate with cross-functional teams on branding, digital products, and business solutions. Automate marketing workflows with n8n to boost operational efficiency, working across packaging, website, and brand logo design.
Built Power BI dashboards to analyze Sales Trends, Inventory Overview, and Customer Flow. Worked alongside data analysts to integrate automation into data processes and reduce manual marketing tasks.
Led visual design for email campaigns and branding projects, collaborating with international clients to deliver high-quality digital campaigns and stronger design systems.
Turning raw numbers into something a business can act on — from cleaning messy data to shipping a dashboard people check every day.
Building models that predict, classify, and support real decisions — evaluated rigorously, not just deployed and hoped for.
Removing the repetitive parts of a workflow with n8n, so teams spend their time on the decisions that actually need a human.
Where it started — brand systems, interfaces, and full-stack builds, carried through everything I make today.
The everyday stack that keeps a data-and-design practice moving without friction.
An automated pipeline that pulls daily wholesale food prices, cleans and stores them in SQL, then flags anomalies and forecasts short-term price movement on an interactive dashboard.
Analyzed 947 recipes to predict which ones drive high site traffic, testing Logistic Regression, SVM, Random Forest, and Neural Nets — landed on Logistic Regression at 84.6% precision, beating the 80% target.
A classification model that predicts loan approval and repayment capacity from applicant data, built to support faster, more consistent lending decisions.
I'm currently open to new opportunities in data science, AI/ML, and product design. Reach out — I read everything myself.