Founding AI Engineer for Industrial Vision-Language Models

CELOTTI TECH TALENT; Singapore; On-site; SGD 180,000 - 300,000; Full time

Job summary

Celotti Tech Talent is seeking a founding AI engineer to build and own industrial vision-language models (VLMs) that interpret engineering drawings, CAD data, and GD&T. You will work directly with the founders and domain experts to develop benchmarks, data pipelines, and training recipes, spanning data, training, evaluation, and serving.

A large GPU cluster is available with no compute budget cap. You will own model development end-to-end, from data collection and labeling to deployment

Qualifications

• Hands-on experience training or fine-tuning vision-language or multimodal models.
• Experience building multimodal training and evaluation datasets.
• Strong ownership, comfort with ambiguity, drive to work on real industrial data.
• Experience with multi-GPU or distributed training.

Responsibilities

• Own the industrial VLM end to end from data and training recipes to benchmarks, deployment, and iteration.
• Shape the multimodal model roadmap with the Chief Scientist.
• Decide when to train, when to fine-tune, and when to combine with OCR or geometry processing.
• Report results with clear metrics and honest failure analysis.
• Collaborate with manufacturing engineers and customers to define what “correct” means on real drawings.
• Publish or open-source selected work to strengthen technical position.

Skills

Python

PyTorch

Multi-GPU training

Deep learning

Education

BS/MS/PhD in CS/AI/EE/ME

Tools

DeepSpeed

FSDP

Hugging Face

Job description

Celotti Tech Talent is seeking a founding AI engineer to build and own industrial vision-language models (VLMs) that interpret engineering drawings, CAD data, and GD&T. You will work directly with the founders and domain experts to develop benchmarks, data pipelines, and training recipes, spanning data, training, evaluation, and serving.

A large GPU cluster is available with no compute budget cap. You will own model development end-to-end, from data collection and labeling to deployment