Amazon’s AGI Information is seeking an exceptional Applied Scientist to drive science advancements in the Amazon Knowledge Graph team (AKG). AKG is re-inventing knowledge graphs for the LLM-era, optimizing for LLM grounding and other LLM-based customer experiences. At the same time, AKG is innovating to utilize LLMs in the knowledge graph construction pipelines to overcome obstacles that traditional technologies could not overcome.
As a member of the AKG Science team, you will have the opportunity to work on interesting problems with immediate customer impact. The team is addressing challenges in automating gap detection, data ingestion, fact verification, query optimization and entity resolution. Your initial focus will be on entity resolution, and you will also have the opportunity to work with scientists working on the other challenges, and with the engineering teams that deliver the science advancements to our customers.
A successful candidate has a strong machine learning and LLM background, is a master of state of the art techniques, has a strong publication record, has a desire to push the envelope in one or more of the above areas, and has a track record of delivering to customers. The ideal candidate enjoys operating in dynamic environments, is self-motivated to take on new challenges, and enjoys working with customers, stakeholders and engineering teams to deliver big customer impact, shipping solutions via rapid experimentation and then iterating on user feedback and interactions.
Key job responsibilities
As an Applied Scientist, you will leverage your technical expertise and experience to demonstrate leadership in tackling large complex problems. You will collaborate with applied scientists and engineers to develop novel algorithms and modeling techniques to build the knowledge graph that delivers fresh factual knowledge to our customers, and that automates the knowledge graph construction pipelines to scale to many billions of facts. Your first responsibility will be to solve entity resolution to enable conflating facts from multiple sources into a single graph entity for each real world entity. You will develop generic solutions that work fo all classes of data in AKG (e.g., people, places, movies, etc.), that cope with sparse, noisy data, that scale to hundreds of millions of entities, and that can handle streaming data. You will define a roadmap to make progress incrementally and you will insist on scientific rigor, leading by example.
