Descripción de la oferta
Descripción del trabajo We are building the operating system for R&D in flavors and fragrances, food and beverage, and cosmetics. A fragrance has 150 to 200 compounds. The data behind it sits in Excel macros, instrument outputs, regulatory PDFs and the memory of a few senior experts. We put an AI data platform inside those workflows, trained on each client's own historical data, and we deploy engineers into the client to do it.Desplácese hacia abajo para encontrar los detalles completos de la oferta de trabajo, incluyendo la experiencia requerida y las funciones y tareas asociadas.Looking for an AI engineer with 2 to 3 years shipping AI software to production.What you will doSit with technicians at the client, understand the workflow and the data, then build it and deliver it.Co-build the platform: ingestion of proprietary instrument formats, spreadsheets and regulatory documents; ERP, LIMS and PLM integrations; Postgres and vector search; knowledge base, agents and ML models on top. Python and AWS. Private instance per client, no cross-client training.Build modules: peak picking, retention time alignment and spectral matching in chromatography; multi-constraint optimization and generative models in formulation; agentic systems over dozens of regulatory sources.10% of your time on long-term research: a paper, or a foundation model trained on chemical data at the Barcelona Supercomputing Center.Own clients and modules. No product manager in between.What we needStrong software engineer first: clean, tested Python, shipping without supervision.Real database work: schemas, query tuning, migrations, data never meant to be machine readable. Postgres expected, vector DBs a plus.LLMs, retrieval, agents and evaluation, with the judgment to know when a model is the wrong tool. Training or fine-tuning is a plus, and so is anything close to chemistry or scientific data.Scientific mindset and high agency. Technical background (maths, physics, engineering, CS, data science) or a credible self-taught profile.Comfortable in front of clients. Spanish and English.Why MaferNobody has trained these models on this kind of data. The layer does not exist yet.Small and very technical: CTO from maths and ML research at MIT, a PhD chemist, a PhD mathematician, engineers from physics and data science. Everyone here is a forward deployed engineer.Working with leading companies in the sector, including one of the largest listed beauty groups. Selected for the BSC AI Factory. $2M pre-seed led by KFund and 4Founders and angels Like Didac Lee, Adrián Mato and Juanjo Mostazo.5k EUR relocation bonus and help finding housing near the office.This is not a 9 to 5 and not a narrow job. Some weeks you are at a client three days in a row, some weeks deep in a training run. You are judged on whether the client's R&D got faster and the platform got better. xkdbapoProcess: 30 minutes with Fer (CEO), technical interview with Marc (CTO), two days working with us at Pier07. We decide in less than three days.
#J-18808-Ljbffr
