Senior Open-Weight LLM Engineer — Remote

EverAI Limited; United States; A distancia; USD 140.000 - 240.000; Jornada completa

Ventajas ofrecidas por este puesto de trabajo

PTO 20 days/year

Annual in-person gathering

Health & wellness stipend

Descripción de la vacante

EverAI Limited is building the future of AI companionship with a globally distributed, fully remote team powering the world's largest AI collaboration platform. You will contribute to open-weight LLMs used by millions, shaping behavior, safety and multilingual capabilities at consumer scale.

The role emphasizes hands-on post-training, fine-tuning with SFT, LoRA/QLoRA, and evaluation, deploying models to production using Python, PyTorch and Hugging Face on GPU clusters in a fast-paced,

Formación

• Post-trained open-weight LLMs for a user-facing product and shipped them to production.
• Experience with fine-tuning is core to the product, not exploratory research.
• Strong Python, PyTorch and Hugging Face with hands-on GPU training.

Responsabilidades

• Fine-tune open-weight LLMs using SFT, LoRA/QLoRA and post-training methods such as DPO and GRPO.
• Diagnose model failures and turn them into experiments; determine data vs. training vs. prompting needs.
• Build data and evals to influence the next training iteration; production deployment.

Conocimientos

Post-training open-weight LLMs

Fine-tuning for product

Python

PyTorch

Hugging Face

GPU training & debugging

Multilingual LLMs

Evaluation & data creation

Moderation & safety

Open-source contributions

Herramientas

PyTorch

Hugging Face

GPUs

Descripción del empleo

EverAI Limited is building the future of AI companionship with a globally distributed, fully remote team powering the world's largest AI collaboration platform. You will contribute to open-weight LLMs used by millions, shaping behavior, safety and multilingual capabilities at consumer scale.

The role emphasizes hands-on post-training, fine-tuning with SFT, LoRA/QLoRA, and evaluation, deploying models to production using Python, PyTorch and Hugging Face on GPU clusters in a fast-paced,