Hackathon Portal
AI Tinkerers - London
Sign in

AI Tinkerers - London Hackathon

Sign in
Machine Learning Engineer: Frontier ML & HPC at a Top London Quant Firm - Dex
Optimise distributed training and frontier accelerators for London quant research. Apply.
Senior Research Engineer: Build Real-Time AI Avatars That Think, Act & React - Dex
Build real-time video diffusion agents at a London unicorn. Apply now.
Engineering Lead at Stealth AI Startup - Stealth AI Startup
Own AI-native architecture and backend pipelines remotely with no equity gimmicks.
All Jobs
Team

FlatScout

Project Concept

Project Concept: Voice-Driven Flat Search Agent

Overview - An agent that helps users find flats on SpareRoom by describing their needs in natural speech. Instead of filling in forms or repeatedly checking listings, the user gives a short spoken brief, and the agent handles the rest, sending them an email with the most relevent new listings every day.

Problem - Flat hunting requires constant manual checking and doesn’t capture how people naturally describe what they want (e.g., “quiet area”, “near a park”, “modern kitchen”). Users end up reviewing many irrelevant listings.

Solution

  1. Voice Briefing – The user speaks a short description of their ideal flat.
  2. Preference Extraction – The system converts this into structured search criteria.
  3. Automated Retrieval – It runs a daily SpareRoom search and collects new listings.
  4. LLM Evaluation – Listings are compared to the user’s preferences, and the best matches are summarised.
  5. Notification – Relevant options are emailed to the user as a simple daily digest.

Why It Fits the Theme
• Voice agent as the primary interface.
• Agentic retrieval performing autonomous daily searching.
• Context management for storing preferences and avoiding duplicates.
• Web retrieval over real property listings.

MVP Scope
• Record a short voice description
• Extract preferences
• Fetch a set of listings via agentic search
• LLM selects and ranks results
• Send an email with the top results

Entry

Status: Submitted

Last saved: December 07 at 6:33 PM GMT

View Entry

Team Roster

Message board not available for this team yet.

Oliver Cairns Team Lead RSVP Approved

data science at lendable
frontend and llm pipeline
I worked as data-scientist for 6 years. recently working at lendable fintech scale-up, where we are interested to apply gen-ai to ops process.
i just want to learn. particularly interested in labelling platforms, evaluation methods, and applications in fintech/regulated industry.
Fyxer AI - founding contractor Lendable - gen-ai data-science team for automation of regulated ops process.

h*********n@g*******m RSVP Waitlisted

n/a
This individual's profile is highly incomplete, lacking any evidence of technical skills, active building, education, or community engagement. Without verifiable artifacts or project summaries, their hands-on coding and shipping experience cannot be confirmed.
no projects mentioned

Robin Haljak RSVP Approved

Data Scientist at Lendable
web scraping and llm pipeline
I am a Cambridge Physicist now working as a Data Scientist at Lendable
Data science, applied machine learning, statistical modelling, feature engineering, experimental design, Python, data engineering, database management
Machine Learning research