Dashboards, ML services, and RAG chatbots for seed-stage SaaS founders. Lean stack defaults. Code your team owns from day one.
RECENT WORK
A fintech SaaS sales team cut per-prospect research from 3 hours to under 1, using a Streamlit dashboard plus ETL I built at Wavess.io.
See the Dashboards serviceA food-tech startup runs demand forecasting and shelf-life prediction on a FastAPI ML service I built at LIFO AI. Multiple retail locations, automated reorder logic.
See the Predictive ML serviceA public LightGBM project on the Credit Card Fraud dataset. 284,807 transactions, severe imbalance. Code, model, and dashboard live on GitHub.
See on GitHubSERVICES
Stop exporting CSVs by hand.
A dashboard sitting on top of your real product data, shipped in 2 to 4 weeks. Streamlit or Metabase, your call. Built on Postgres so your team isn't paying per-seat fees forever.
Automate the decision someone makes 100 times a week.
Forecasting, scoring, anomaly detection. Deployed as a FastAPI service with monitoring and a retraining script your engineers can run themselves.
Let your team query your docs in plain English.
A RAG chatbot over your internal docs, tickets, or playbooks. Vector RAG on Postgres pgvector for most cases. GraphRAG with Neo4j when your data has relationships worth following. Pilot pricing on the first three engagements.
How It Works
A 30-minute call. We go through your data setup and the decision you want the data to inform. I tell you straight whether the project is a fit for me. No deck. No pitch.
I write up a discovery document covering the build plan, the stack, milestones, and your fixed price. If at the end of the week it isn't the right fit, you keep the document and we part ways. No contract trap.
I work in your repo. I push to staging. We have one weekly video call. You see progress every week, not at the end.
Code review with your engineers. Runbook for retraining, redeployment, and monitoring. After this week, your team owns the system.
The setup: A fintech SaaS startup's sales team was spending 3 to 4 hours per prospect, stitching Trustpilot reviews, hiring data, and industry news together by hand. Some pitches got prepped poorly. Some got skipped.
What got built: An ETL pipeline that pulled 520+ data points (154 hiring records, 366 customer reviews) from Trustpilot API and industry news scrapers. Multilingual sentiment analysis with text normalization for regional languages. A Streamlit dashboard with filters by competitor, market segment, and time period.
What changed: Per-prospect research time dropped 70%, from 3 hours to under 1. Reps walked into pitches with named competitor weaknesses to position against. The team ran more pitches per week with better prep on each one.
Stack: Python, n8n, BeautifulSoup, Trustpilot API, Streamlit.
Work done as Data Science Intern at Wavess.io between July and October 2025.
WHO YOU'RE HIRING
A full-time freelance data scientist based in Mumbai, India. Bachelor's in AI and Data Science from Mumbai University. Two startup engagements behind me: a fintech competitive-intelligence dashboard at Wavess.io and an inventory ML service at LIFO AI.
The case studies on this site are anonymized but real. The GitHub link in the nav has the public-data versions if you want to read the code first.
From the Blog
us enim, eget posuere nulla tellus quis orci. Aliquam auctor aliquam magna, ac interdum tellus condimentum non. Ut nulla lorem, porta non pharetra nec, maximus fringilla eros. Maecenas…
Read posturabitur ut dignissim augue. Vivamus consequat odio a tempor egestas. Vivamus nulla nulla, varius nec vehicula nec, malesuada…
Read postWelcome to WordPress. This is your first post. Edit or delete it, then start writing!
Read postOne of three projects: a dashboard on top of your scattered product data, a forecasting or scoring model deployed inside your product, or a RAG chatbot that answers questions over your internal docs. The Services pages have specifics.
I quote per project. You get one fixed number after a 20-minute scoping call. Pricing is calibrated to seed-stage budgets and well below typical agency or senior-freelancer rates. If scope changes mid-project, we agree to a written change order before anything extra gets built.
Lead time is one to two weeks. Engagements run 4 to 12 weeks. I take a maximum of two clients at a time so each project gets focus.
Yes, fully remote. I worked async with German product and sales teams at Wavess.io and the rhythm worked. Slack, Notion, GitHub, plus a weekly video call covers most needs.
Yes. That's what the scoping call is for. Most founders come in asking for one and leave with a different recommendation. The first call is free and there's no obligation after.
Python, FastAPI, PostgreSQL on the backend. Streamlit and Metabase for dashboards. LangChain, LlamaIndex, and pgvector for RAG. Neo4j and Ollama for graph-based RAG. n8n for workflow automation. Power BI or Tableau if you have an existing license.
Yes. About half of my work is translating between business questions and technical implementation. The Wavess.io dashboard was used by a sales team that didn't write SQL. The output was a one-click dashboard, not a query tool.
Me. There's no junior to hand the work to and no offshore team behind the scenes. I take a maximum of two engagements at a time so each one gets full attention.