01Home 02Research 03Teaching 04Students 05Products 06News 07Writing 08Platform New 09Contact 10Curriculum Vitae
Web Products

Research-grade thinking, shipped as software.

Beyond the lab, I design and build full-stack web products that put data, machine learning and clean UX in front of real users. Here are three I've shipped recently — a property-intelligence dashboard, a postgraduate study planner for New Zealand, and an ML-driven flight-fare advisor — each solving a concrete, everyday problem.

Open HouseScout ↗ Open PathFinder ↗ GitHub ↗
3Products in portfolio
2Live & publicly usable
AI + MLLocal LLM & predictive models
Full-stackPython & JavaScript
01 — Portfolio

Products I've designed & built

Live Property Intelligence Python · Local AI

HouseScout

A property-buying dashboard that helps Christchurch home buyers find affordable homes that match very specific criteria — and stay on top of the market without endlessly refreshing listing sites. Built for investment-minded buyers who want to "live in & rent rooms to pay it off fast."

Key features

  • Listings tracker monitoring user-defined criteria
  • Central dashboard with key metrics at a glance
  • Searchable, filterable listings browser
  • Indicative property valuations & estimates
  • Market insights & analytics on matches
  • Buying-journey planner
  • Private local AI chat (powered by Gemma)
  • Personalised settings & saved preferences

The value: turns a scattered, stressful search into a focused, data-driven workflow — surfacing sub-$500k homes with the right features (garage, backyard) so first-home buyers and investors can move fast and buy with confidence. Estimates are indicative, not registered valuations — and not financial advice.

Open Live App ↗ Source on GitHub ↗
ureshan2011.github.io/HouseScout
HouseScout dashboard — a Christchurch buying dashboard showing 523 listings tracked, 10 matching the buyer’s criteria, an average matching price of $461,500, a valuation-model confidence badge, and scored property recommendations with price, suburb, bedrooms, garage and rooms-to-rent tags
Live Study Planning · New Zealand Static Web · Client-side

PathFinder

Applying for postgraduate study abroad means piecing the answer together from a dozen half-reliable sources — agent blogs, forum threads, PDFs that are three years out of date. PathFinder puts the whole route in one place for Sri Lankan students heading to New Zealand: what you qualify for, what it really costs, the visa, and the move.

Key features

  • Seven-stage tracker, assessment to first week
  • Five-minute assessment with a personal timeline
  • 1,716 NZQF level 8 & 9 qualifications, searchable
  • 51 providers across 11 subject areas
  • Cost calculator in NZ$ and rupees
  • Visa funds, deadlines & settlement guidance
  • 281 scholarships and funding sources
  • Mentors already studying in New Zealand

The value: replaces guesswork and agent fees with the published record — real entry requirements from the NZQA register, real fees, real visa thresholds — so a student can see in five minutes whether the route is open to them, and what the next month should look like.

Open Live App ↗ Source on GitHub ↗

Free to start, no account needed. PathFinder is an independent tool built from published sources — it is not affiliated with NZQA, Immigration New Zealand or any provider, and it is not immigration advice.

ureshan2011.github.io/Pathfinder
PathFinder programme catalogue — “Find a master’s programme worth applying to”, with 1,716 current NZQF level 8 and 9 qualifications searchable by subject, provider and level, subject filters showing counts per field, and result cards for master’s programmes at Massey University and Victoria University of Wellington
Live Travel Python · Machine Learning

Faro

A flight-fare timing assistant that answers the one question every traveller has: book now, or wait? Faro tracks prices across airlines and uses a model trained on route-specific history to give a clear, daily Book / Wait call — removing the guesswork (and regret) from buying tickets.

Key features

  • Daily fare monitoring across airlines
  • Clear "Book now" vs "Wait" recommendation
  • ML trained on route-specific price history
  • Full flight details — airline, stops, times
  • Refreshes several times a day, auto-reloads
  • Aggregates fares from Google Flights
  • Live weather & currency context built in
  • Open, publicly accessible underlying data

The value: flexible travellers save money by booking in the optimal window — guided by predictive modelling instead of gut feel — while a single dashboard replaces hours of manually re-checking fares.

faro · fare timing
CHC → CMB Christchurch to Colombo · 14 Nov · 1 adult BOOK NOW 87% confidence TODAY’S BEST FARE NZ$1,284 ↓ 11% wk 30-day low NZ$1,241 30-day high NZ$1,690 Fare history · last 30 days 1700 1500 1300 today 30d ago SQ Singapore Air 1 stop · 17h 40m $1,284 MH Malaysia Air 1 stop · 19h 05m $1,347
02 — Approach

How I take an idea to a shipped product

The same rigour I bring to research — clear problem framing, evidence, and evaluation — applied to building software people genuinely use.

Full-stack web

End-to-end builds in Python and TypeScript — from data layer to a clean, responsive interface.

AI & machine learning

Predictive models and local LLMs wired into products — practical AI that earns its place in the UI.

Data pipelines

Aggregating and cleaning real-world data — listings, fares, weather, currency — into something dependable.

Product UX

Designing focused interfaces that make a complex decision feel simple and trustworthy.

FAQ

Questions about these products

Who built HouseScout, PathFinder and Faro?

All three products were designed and built by Dr. Yasas Sri Wickramasinghe — a researcher, university lecturer and software developer based in Christchurch, New Zealand.

Are HouseScout and PathFinder free to use?

Yes. HouseScout and PathFinder are both live and free to use directly in your browser, with no account needed to start. Faro is a personal build and is not publicly hosted.

What technologies are these products built with?

HouseScout and Faro are built with Python, with Faro also using a machine-learning model for price prediction. PathFinder is a static browser application with no backend — the assessment, the cost calculator and the NZQA programme catalogue all run client-side.

What does HouseScout do?

HouseScout is a property-buying dashboard for Christchurch home buyers. It tracks listings against your criteria, surfaces indicative valuations and market insights, includes a buying-journey planner, and offers a private local AI assistant powered by the Gemma model. Estimates are indicative, not registered valuations — and not financial advice.

How does Faro decide whether to book or wait?

Faro uses a machine-learning model trained on route-specific historical fares to predict how a price is likely to move, then gives a clear “Book now” or “Wait” recommendation. It monitors fares across airlines and refreshes several times a day.

Who is PathFinder for?

Sri Lankan students planning a master's or PhD in New Zealand. It covers the whole route in one place — a five-minute readiness assessment, a searchable catalogue of 1,716 current NZQF level 8 and 9 qualifications from 51 providers, real tuition, living and visa-funds costs in NZ$ and rupees, 281 scholarships, and mentors already studying in New Zealand.

Where can prospective MBI students learn more?

See my guide for prospective students for MBI programme facts, my teaching approach, and a free consultation link. It's my own account as the lecturer who teaches the programme, not an official Yoobee College page.

Let's build

Have a product idea that needs data, AI or great UX?

Get in Touch → See more on GitHub ↗