Paris · hace 2 semanas
Diseñarás, entrenarás y desplegarás modelos de ranking y recuperación para agentes de IA, construirás marcos de evaluación para medir la calidad de búsqueda y optimizarás la latencia y el coste de GPU del servicio de modelos.
About the role Linkup is the web search engine built for AI. Our infrastructure crawls, indexes and serves the live web to AI agents, and it’s scaling fast. We’re looking for a top‑tier Machine Learning Engineer to join Linkup and work hand‑in‑hand with our CTO, Denis Charrier, and the rest of the tech team. Your core responsibilities will be to: Design, train and ship ranking and retrieval models for AI agents • Design, train and ship ranking and retrieval models for AI agents • Bring new ranking signals into the engine, beyond pure relevance • Build the evaluation framework that tells us what really improves search quality • Optimize model serving for latency and GPU cost • Turn customer and benchmark feedback into ranking improvements, with clear before/after numbers Requirements You’re a great match if you: • Are excited to help build Google for AI agents- and believe in Linkup’s mission • Have strong hands-on experience in information retrieval, ranking or search in production. • Build rigorous evals and check the data behind every number • Care about latency, cost and reliability as much as model quality • Thrive in a fast-paced, high-ownership startup environment Example Projects • Ship a new ranking model to production and prove the gain on our eval sets • Push our models to the next level: better answers, lower latency, lower cost per query • Build and maintain the evaluations framework that measures every change to our search engine Our Stack Python, PyTorch, Hugging Face, vLLM, Ray, Spark / Databricks, Vespa, Kubernetes, GitHub Actions, Terraform Details The ideal candidate will play a central role in our engineering team, partnering closely with the founders. The models you ship will decide what thousands of AI products read every day. We offer a flexible remote‑work policy- see the details here. Join us and leverage your expertise to push the boundaries of AI! About Linkup Our Vision We believe in a future of AI agents and the agentic web. These digital entities can create more value for humanity than we ever could alone, but they depend on the internet to operate, collaborate, take action, and gather information. AI agents need web search. To perform well, they require information that is current, reliable, and context‑specific—not the diffuse, steadily aging knowledge frozen at their last training cut‑off. Modern web search is built for humans, not AIs. Instead of trusted facts, it serves links, ads, and click‑bait. People want visually engaging pages; AI systems need structured truth. An AI can absorb oceans of data, while human attention barely skims the surface. Humans think visually and intuitively; AI reasons in latent space. Web 2.0 is sunsetting. The deck is being reshuffled, and the Bing‑Google search oligopoly is crumbling. Indexing ever more pages is no longer the goal; semantically organizing facts for AI systems is. In the coming agentic web, Google’s ad‑driven model has no natural home. Search for AI must be rebuilt from the ground up. We are reconstructing the web as a rigorously structured knowledge layer, mapping every concept, entity, and claim into a coherent semantic graph. This foundation delivers verified facts and rich context to AI instantly, unlocking faster reasoning, higher accuracy, and radically more useful answers for everyone. Humans search Google; AI searches Linkup. Our Mission Reality never stands still, and neither should intelligence. Linkup reconnects language models to a living stream of verified facts, ensuring AIs remain as current and insightful as the world they serve. Our mission is to deliver the world’s freshest, most trusted knowledge to AIs. Our Culture Our culture is built on the following values: • Mission. We believe we play a critical role in building the agentic web. We are 100% committed to our mission of bringing the reality of the world to AI agents. • Speed. We build fast, we build to last. We get things done, well. • Ownership. We empower every team member to take ownership of their projects, ensuring they see them through to successful completion. • Teamwork. Our strength lies in our unity. We rely on one another and embrace our diversity as a core asset. • Humility. We maintain low egos, showing respect for others and continuously learning from their insights and experiences. We expect all team member to take an active role in shaping our company culture. Our Team We are a small technical team, working shoulder to shoulder to build the web of agents. A bit more about our 3 co-founders: • Philippe Mizrahi- Led the development of Lyft’s Autonomous Vehicles marketplace. Previously launched Amundsen, an open source search engine. • Denis Charrier- Founding Engineer at Niland, an audio search company (acquired by Spotify), then Staff Engineer at Spotify. 15y of software development experience. • Boris Toledano- Led robotics at Carrefour, after 3y at McKinsey & Co working on public affairs and digital transformation.
Publicada el 25 oct 2026
