Software Engineer, Agentic Matching (Python / GCP)

About Ayo

Ayo is building an AI native way for people to find each other. Someone shows up with an intent. "I just moved to Lisbon and want to make friends who like hiking." "I want to practice Spanish with a native speaker over coffee." "I'm flying to Athens next week for a conference and want to meet people in the same field." Ayo understands that intent and matches them with the right person. The intent can be anything, and our job is to make sense of it and find the human on the other side.

Two things make this hard and worth doing:

1. An AI agent that talks to real people to draw out and encode what they actually want. It works over a dedicated app and channels like WhatsApp and Telegram, where users simply add Ayo to their contacts like any other person.

2. A matching engine that turns open ended intents into real person to person matches.

We are small, moving fast, and looking for an engineer to help build this across the whole stack.

The interesting problems you'll work on

- Open ended intent. People express what they want in messy, human language. Turning that into something a machine can match on, and knowing when to ask a clarifying question instead of guessing, is the core of the product.

- Two sided quality. A good match has to work for both people. That is a ranking and reasoning problem, not just nearest neighbor lookup.

- Cold start and liquidity. Matching is only as good as the pool of people available. We need to make early users feel matched well even when the network is still small.

- Latency and cost. LLM calls are powerful and expensive. Deciding when to use a cheap embedding lookup versus an LLM reasoning step, and keeping the whole thing fast, is a real engineering constraint.

- Trust and safety. We are introducing strangers to each other, so quality, filtering, and abuse prevention matter from day one.

What you'll do

- Build and improve the conversational agent that collects, clarifies, and encodes user intent across app, WhatsApp, and Telegram.

- Design and iterate on the matching engine: embedding lookups, vector search, LLM based ranking and reasoning, and the logic that combines them.

- Own features from the messaging integration, through the backend services, to the data and model layer.

- Instrument and evaluate match quality so we can improve it with evidence rather than vibes.

- Set up scalable, secure infrastructure on GCP to run agents and models reliably in production.

- Work directly with the founders on product decisions and shape how matching actually works.

What we're looking for

- Strong Python engineer who can own systems from front to back.

- Hands on experience building with LLMs: prompting, tool and function calling, RAG, and working with embeddings and vector search.

- Comfortable on GCP (Cloud Run, Pub/Sub, Cloud SQL, GCS, or close equivalents) and shipping to production.

- Product sense. You can turn a fuzzy human need into a working system and iterate based on real usage.

- Comfortable with ambiguity and a fast moving early stage environment.

Nice to have

- Experience integrating WhatsApp or Telegram (Business API, bots, webhooks).

- Background in recommendation, matching, search, or ranking systems.

- Familiarity with agent frameworks and orchestration.

- Experience running LLM or ML workloads cost effectively at scale.

Our stack

Everything is Python and runs on GCP. We use LLM APIs, embedding models, and vector search, with messaging integrations for WhatsApp and Telegram.

Why Ayo

- A genuinely novel product at the intersection of AI agents and human connection.

- Real ownership as an early engineer, not a cog in a large machine.

- A short path from idea to production, working directly with founders.

- A senior team that has done this before. The people behind AYO have built and sold AI businesses at scale.

- Competitive Salary, remote working and the potential for options.