Senior Generative AI Research Engineer (Foundation Models) — Tokyo

Location requirement

You must already be living in Japan with valid work authorization. We are not sponsoring relocation or visa transfers from overseas for this role. Applications from outside Japan will not be considered.

About Abbeal

Abbeal is an international IT consulting firm (France, Japan, Canada) placing senior engineers on high-impact R&D teams. Through Abbeal, build a custom foundation model for sustainable cosmetics & pharma R&D with a Tokyo-listed group.

Context

You will build a custom foundation model that generates sustainable cosmetic & pharmaceutical formulations in unsupervised mode, for a Tokyo-listed pharma & cosmetics group. Deep generative AI / foundation-model research, remote-friendly (Tokyo / Paris / remote). English is the working language; Japanese is not required (no JLPT requirement).

Your Mission

Design and train custom foundation / generative models for molecular & formulation generation

Work with GNNs, diffusion models, and unsupervised learning on chemistry data

Leverage RDKit / DeepChem and modern deep-learning tooling (PyTorch / JAX)

Build RAG and evaluation pipelines around the models

Collaborate with R&D scientists to translate research into usable outputs

Profile

Currently based in Japan with valid work authorization — no relocation or visa sponsorship from overseas for this role

5+ years (ideally 8–12) in ML / deep learning, with generative or foundation-model experience

Strong PyTorch / JAX; GNN, diffusion, unsupervised learning

Familiarity with cheminformatics (RDKit / DeepChem) a strong plus

Fluent English; Japanese not required

Location and Setup

Remote-friendly (Tokyo / Paris / remote). Freelance (senior FR/EU day rate) or permanent in Japan via Abbeal KK. Hired and supported by Abbeal (visa sponsorship + Mobbeal expat package). Indicative permanent range ¥14M–¥20M/year depending on experience.