Applied LLM Fine-Tuning Engineer

Visa Hunt; San Francisco, Northern (CA, KY); Hybrid; USD 140,000 - 210,000; Full time

Benefits offered by this job

Top-tier compensation

Stock options

Health & wellness

Meals provided

Parental leave

Unlimited vacation (US)

Visa sponsorship

Team events

Job summary

Reflection is seeking a ML Engineer to join the Applied AI team to fine-tune open-weight models for enterprise customers. You will work hands-on with customer data, run training loops, build evaluation harnesses, and help deploy adapted models to production across hybrid environments.

You will collaborate with research teams, design evals, and contribute to playbooks. The role emphasizes strong Python skills, data pipelines, and experience shipping models to production in startup settings.

Qualifications

• Applied ML experience with hands-on fine-tuning of language models.
• Familiarity with SFT, DPO, RLHF, or similar techniques.
• Experience building data pipelines and evaluation benchmarks.

Responsibilities

• Fine-tune open-weight models for customer-specific use cases.
• Build and maintain evaluation infrastructure and baselines.
• Prepare training data from raw customer inputs and ensure data quality.
• Debug training and inference issues and monitor loss curves.
• Support end-to-end deployments across cloud, VPC, and on-premises.
• Contribute to playbooks and best practices for fine-tuning and evals.

Skills

Fine-tune LMs

SFT training

RLHF

Python

Data pipelines

Version control

Evaluation design

Tools

GPUs

PyTorch

TensorFlow

Job description

Reflection is seeking a ML Engineer to join the Applied AI team to fine-tune open-weight models for enterprise customers. You will work hands-on with customer data, run training loops, build evaluation harnesses, and help deploy adapted models to production across hybrid environments.

You will collaborate with research teams, design evals, and contribute to playbooks. The role emphasizes strong Python skills, data pipelines, and experience shipping models to production in startup settings.