ML & backend for an ambitious Al-Powered Hotel Search Engine (Astra)

Astra:

Name code for a new ambitious project of building an alternative to Booking.com powered by Al.

Led by a team of seasoned founders (Tiller acquired by SumUp, Zenly acquired by Snapchat)

Mission:

Build and evolve the backend that powers Astra—an Al-driven search engine for hotels.

You'll work on transforming natural-language queries into structured searches, managing hotel data,

orchestrating Al calls, and integrating external sources.

Key Responsibilities

Design and maintain search logic: Handle full flow from user query → Al parsing → DB filtering → ranked hotel results.

Session & context management: Persist user search sessions, filters, and chat context across multiple screens (Home, Search, Hotel).

Integrate LLM reasoning: Orchestrate OpenAl API calls (or equivalent) for query understanding, re-ranking, and contextual responses.

Database management: Maintain and optimize Supabase (PostgreSQL) schemas for hotels, amenities, reviews, and user sessions.

External data enrichment: Connect and normalize external APIs (eg. Booking.com, Google Places, Tavily/Serper) for complementary data.

Performance optimization: Ensure low-latency search queries and scalable architecture (caching, pagination, indexing).

Required Skills

Core stack: Python (FastAPI), Supabase and PostgreSQL.

API design: Strong experience with REST APIs (GraphQL is a plus).

Al integration: Experience consuming LLM APIs (OpenAl, Anthropic, etc.), using RAG techniques and handling structured outputs (JSON schema).

Data modeling: Ability to design flexible hotel/search schemas and manage complex filters.

Session management: Familiar with persistent user context and conversational states.

Query interpretation: Ability to map natural-language queries into structured search parameters.

DevOps basics: Familiar with modern CI/CD, Docker, and deployment on Vercel, Railway, or Supabase Edge Functions.

Nice to Have

Knowledge of semantic search / embeddings / vector DBs (Weaviate, Pinecone, etc.).

Familiarity with Al orchestration tools (LangChain, LlamaIndex, n8n).

Understanding of prompt engineering and LLM function calling.

Basic frontend awareness (React/Next.js) to collaborate closely with Ul developers.

Experience in travel or marketplace data models (availability, amenities, geolocation).

Profile

10+ years backend experience, incl in startups and/or projects from scratch

Product-minded engineer who can balance speed and structure.

Curious about Al and conversational interfaces.

Comfortable iterating fast in a small, design-driven and fast-pace startup environment.