Machine Learning Engineer at Hostelworld Group

Hostelworld Group; Porto; Presencial; EUR 50 000 - 80 000; Tempo integral

Vantagens oferecidas por esta oferta de emprego

Competitive salary & benefits

Enhanced annual leave + wellbeing days

Paid family leave

Agile working (Working from Abroad)

Growth & development support

Inclusive people policies

5 paid volunteering days

Resumo da oferta

Hostelworld Group is seeking a Machine Learning Engineer in Porto to build production ML/LLM systems that power Hostelworld discovery and recommendations. You will work on MCP servers, agent-facing tools, and LLM applications, focusing on improving accuracy, latency, and cost.

The role involves model evaluation, experimentation, and collaboration with Product, Engineering and Analytics teams to deliver impactful AI features and reliable services. Hybrid/onsite options in Porto.

Qualificações

• Proficient Python and SQL with clean, testable code.

Responsabilidades

• Build production ML and LLM systems for Hostelworld discovery, relevance and efficiency.
• Design, implement and operate MCP servers and agent-facing APIs.
• Develop LLM applications using RAG, embeddings, vector search and structured outputs.
• Improve model accuracy via benchmarks, error analysis, prompts, fine-tuning and distillation.
• Evaluate open-source or commercial models for quality, latency and cost.
• Own end-to-end productionisation: data pipelines, embeddings/features, serving, monitoring and docu‑mentation.

Conhecimentos

Python

SQL

Production ML

LLM engineering

ML services

LangChain

LlamaIndex

Model evaluation

Model trade-offs

Cloud & ML infra

Communication

Ferramentas

GCP

BigQuery

Vertex AI

Kubernetes

Redis

PostgreSQL

MongoDB

LangChain

LlamaIndex

Descrição da oferta de emprego

Apply for Machine Learning Engineer at Hostelworld Group in Porto, P13, PT. This Full time on site position offers great opportunities for career growth.

About Us

Hostelworld is the world's leading Social Travel platform – a social network‑powered OTA on a mission to help travellers find their people and create unforgettable moments together. Founded in 1999 and headquartered in Ireland, we are a team of around 270 people serving hostel partners across more than 180 countries. Since launching our social network in 2022, we've welcomed over 3.4 million social members who aren't just booking trips, they're messaging, finding events, meeting up and becoming powerful brand advocates. This community generates a rich proprietary dataset of 16 million+ chat messages and 17 million bookings, powering AI‑driven personalisation and network effects that no generalist OTA can replicate. We are not just another OTA – we created an entirely new category – Social Travel.

Our Culture

At Hostelworld, our culture is a direct reflection of our customers: adventurous, curious, and social. We have a shared love of travel that fuels our work and connects us on a deeper level. It’s a fast‑paced environment that blends the agility and "scrappy" resourcefulness of a start‑up with the experience and ambition of a global, publicly listed company. We are a team of pragmatic optimists who are data‑obsessed and results‑driven. We value a "test and learn" mindset, encouraging experimentation and empowering our teams to take calculated risks. We believe in doing the right thing—for our customers, our partners, and each other. We foster a supportive and collaborative atmosphere where diverse perspectives are celebrated, and where every team member has the opportunity to make a significant impact on our journey. We embrace the journey, not just the destination, and we seek leaders who will thrive in a dynamic environment and inspire their teams to do the same.

Location

This role is based in Portugal. We have an office hub in Porto available for those who prefer a hybrid model where you can spend time with colleagues in‑person. You will need to be able to commute to our office hub as required from time to time.

What the Role Looks Like

Build production ML and LLM systems that improve Hostelworld discovery, relevance and efficiency. This role focuses on MCP servers, agent‑facing tools, LLM applications, model evaluation, and adapting open‑source or commercial models to improve accuracy, latency and cost.

What You’ll Do

• Translate traveller, hostel and business problems into ML/LLM solutions across search, ranking, destination discovery, property Q&A, social matching and content understanding.
• Design, build and operate MCP servers and agent‑facing APIs that expose Hostelworld capabilities safely to LLMs and external agent ecosystems.
• Build LLM applications using RAG, embeddings, vector search, tool/function calling, structured outputs, moderation and product logic.
• Improve Hostelworld‑specific accuracy through benchmarks, error analysis, prompt/version control, fine‑tuning, distillation, synthetic data and model selection.
• Evaluate and adapt open‑source or commercial models where they improve quality, latency, vendor flexibility or inference cost.
• Own productionisation end‑to‑end: data pipelines, embeddings/features, model serving, APIs, monitoring, alerting, documentation and support.
• Work with AI Gateway, data platform and engineering teams on model routing, caching, rate limits, cost observability, guardrails and provider abstraction.
• Define hypotheses, run experiments and communicate technical, product and commercial impact to Product, Engineering, Analytics, Marketing and senior stakeholders.

Key Focus Area

• MCP and agent visibility: hostel search, property Q&A, city/social context, deep links and logging to identify blind spots in AI‑mediated travel discovery.
• Hostelworld‑specific LLM quality: evaluation datasets using inventory, reviews, attributes, social signals, booking context and destination data.
• Cost and latency optimisation: model routing, caching, smaller models, batching, quantisation, distillation and provider selection.
• Reusable infrastructure: evaluation harnesses, deployment patterns, libraries and templates for a small high‑expertise ML team.

What We’re Looking For (Must Have)

• Strong Python and SQL, with modular design, testing, CI/CD, code review, documentation and operational ownership.
• Production ML experience, ideally in search, ranking, recommendations, personalisation, marketplaces, travel, e‑commerce or consumer products.
• Practical LLM engineering: RAG, embeddings, vector search, tool/function calling, structured outputs, prompt management and monitoring.
• Experience building ML/LLM services. MCP servers, agentic workflows, LangGraph/LangChain/LlamaIndex or equivalent orchestration frameworks.
• Strong evaluation mindset: offline benchmarks, relevance evaluation, A/B testing, error analysis, factuality/safety checks, latency and cost measurement.
• Ability to reason about model trade‑offs: frontier vs small model, fine‑tuned vs prompted, traditional ML vs rules, and quality vs latency vs cost.
• Working knowledge of cloud/data/ML infrastructure such as GCP, BigQuery, Vertex AI, Kubernetes preferred; Redis, Postgres, MongoDB/vector stores or similar also applicable.
• Clear communication with engineers, product managers and senior stakeholders.

Nice to Have

• Fine‑tuning, distillation, quantisation or serving open‑source models.
• Agentic experience.
• Multilingual travel content, review summarisation, moderation, social graphs, recommendation systems or marketplace ranking.
• Synthetic data or prompt‑set generation for AI visibility, competitor benchmarking or agent‑behaviour measurement.

Success Profile

• First 90 days: understand Hostelworld ranking, social travel, Q&A, AI Gateway and MCP priorities; ship or materially improve one LLM/MCP component or evaluation framework.
• Six months: own a measurable workstream improving LLM answer quality, agent discoverability, latency or cost with monitoring and product/business impact.
• Long term: become a go‑to technical owner for Hostelworld’s agentic AI and LLM infrastructure.

What We Offer

• Competitive salary & benefits
• Enhanced annual leave plus 3 Wellbeing Days per year
• Paid family leave (maternity, paternity, surrogacy & adoption)
• Agile working (plus a Working from Abroad Policy)
• Support for your ongoing growth & development
• Inclusive people policies (sickness, menopause, compassionate and fertility leave)
• A chance to give back to your local community with 5 paid volunteering days

Our Behaviours

• Grow others – We fundamentally believe that investing in growing others benefits everyone.
• Master it – We are obsessed with our area of expertise and enjoy developing our skills.
• Collaborate – We are in it together, for the tough stuff and the celebrations too.
• Adapt – We work fluidly, adapting to new information and the evolving environment while staying committed to our goals.
• Innovation and experimentation fuel our projects and we’re never afraid to pivot.
• Deliver – We focus on the end result; we value outcomes over activity.

We lay out the above requirements to guide applicants to the experiences that we believe will allow you to be successful in the role. If you don’t meet them all, please consider applying if you think you can still perform the role as described.