This role may have closed.
Listings older than 30 days are deprioritized in search. The link to the source is still below in case the posting is still open. For active openings, see similar jobs.
Upstage
AI Research Engineer - Post-training (VLM Tech Lead)
Seoul, South KoreaPosted 1 month ago
What you'd do
- Lead VLM post-training data pipeline covering collection,refinement,and generation
- Optimize visual reasoning performance of vision-language models
- Collaborate closely with Central Training team on model training and testing
What they want
- VLM research publications (1st or corresponding author at top ML/NLP conferences) or equivalent VLM product commercialization experience
- Experience building high-quality data pipelines for VLM/LLM that achieved state-of-the-art results
- Broad deep understanding of multimodal and foundation model development trends
Nice to have
- Agent Harness design experience for UI/UX and screen-recognition data synthesis
- Experience improving VLM post-training pipelines via RLVR or OPD
- Kaggle or international AI competition high rankings
Seoulstart's read
- Visa tier (likely)
- E-7 specialty / professional (likely)
- Korean language
- Native Korean fluency required
- English friendliness
- Medium: English JD, mixed-language workplace