Remote LLM Research Scientist - Pre/Post-Training

Synthires; California; Remote; USD 138,000 - 165,000; Full time

Benefits offered by this job

Weekly payments

Job summary

Synthires is seeking an LLM Research Scientist (Pre-training & Post-Training) to contribute to cutting-edge AI research in a remote, hourly-contract role. The candidate will lead pre-training, post-training, data curation, and evaluation for large language models, and collaborate with researchers to improve alignment, efficiency, and performance.

Experience with RLHF, SFT, and scalable training pipelines is a plus; a PhD or equivalent research track is highly valued.

Qualifications

• 3+ years of machine learning research experience, PhD qualifies.
• Strong expertise in LLM post-training, RL for LLMs, model alignment, and evaluation.
• Experience with PyTorch, JAX, TensorFlow or similar frameworks; transformer basics understood.
• Deep understanding of transformers, LLMs, optimization and modern deep learning methods.
• Excellent analytical and scientific communication skills; remote work capability.

Responsibilities

• Train transformer-based language models from scratch and fine-tune open-weight foundations.
• Design pre-training and post-training pipelines for large language models.
• Create, curate and optimize large-scale training datasets from diverse sources.
• Develop data filtering, deduplication, quality classification, and curriculum strategies.
• Diagnose and resolve training failures, convergence issues and instabilities.
• Build and evaluate SFT, DPO/RLHF/RLAIF, and reward modeling pipelines.
• Design evaluation benchmarks and analyze model performance with rigorous experiments.
• Collaborate with AI researchers to improve reasoning, alignment, efficiency and performance.

Skills

3+ years of ML research

LLM Post-Training

Reinforcement Learning for LLMs

Model Alignment

LLM Evaluation

Transformer architectures

PyTorch/JAX/TensorFlow

Education

PhD in Computer Science / ML / AI / NLP

Top-100 university / FAANG background

Tools

PyTorch

JAX

TensorFlow

Job description

Synthires is seeking an LLM Research Scientist (Pre-training & Post-Training) to contribute to cutting-edge AI research in a remote, hourly-contract role. The candidate will lead pre-training, post-training, data curation, and evaluation for large language models, and collaborate with researchers to improve alignment, efficiency, and performance.

Experience with RLHF, SFT, and scalable training pipelines is a plus; a PhD or equivalent research track is highly valued.