LLM Fine-Tuning Engineer

About the Company

Bright Vision Technologies is a software company focused on creating modern solutions that help organizations streamline and improve their operations. The team builds scalable, secure, and easy-to-use applications using advanced technology.

The company is hiring an LLM Fine-Tuning Engineer to help advance its mission of improving business processes through intelligent technology. This is a strong opportunity for someone looking to grow within a respected organization with significant long-term potential.

Role Overview

This position is a 100% remote role for candidates located in the Continental United States. It is an in-house Statement of Work engagement with Bright Vision Technologies, where Bright Vision Technologies is the client, end customer, and employer. The role is full-time and follows a direct W2 employment model.

The engagement is intended to be long-term and multi-year, aligned with the company’s SOW delivery roadmap. A technical coding assessment is required for all applicants.

Compensation and Employment Terms

The salary range for this position is $100,000 to $150,000 per year, with compensation based on experience and accompanied by benefits.

This role does not support C2C, 1099, or third-party staffing arrangements. Candidates must be willing to work directly as W2 employees of Bright Vision Technologies.

No new H1B sponsorship is available. However, qualified candidates currently on H1B may apply for transfer support.

Job Summary

The LLM Fine-Tuning Engineer will be responsible for designing, running, and operationalizing fine-tuning workflows for large language models using supervised learning, preference optimization, and reinforcement learning methods. The role demands strong practical knowledge of modern training stacks, disciplined dataset preparation, robust evaluation methods, and the ability to keep complex training pipelines running reliably.

The ideal candidate combines ML depth with production engineering habits, understands the trade-offs between data quality, compute usage, evaluation standards, and delivery speed, and can work closely with product, design, engineering, operations, and business partners to turn ambiguous needs into dependable solutions. The role also includes code review, design review, and mentorship responsibilities.

Key Responsibilities

• Plan and run fine-tuning experiments for large language models using supervised learning, DPO, RLHF, and similar methods.
• Lead the creation, curation, and quality control of instruction and preference datasets.
• Develop training pipelines that can scale using modern distributed training frameworks.
• Adjust hyperparameters, optimizer settings, and training stability methods for large-model tuning.
• Apply parameter-efficient tuning approaches such as LoRA, QLoRA, and adapter-based techniques.
• Build thorough evaluation frameworks using automated tests, human review, and task-specific probes.
• Set up safety, refusal, and policy checks to monitor model behavior across releases.
• Run large training workloads on GPU clusters and troubleshoot failures while preserving training progress.
• Improve training efficiency through mixed precision, sequence packing, and optimized attention methods.
• Track model artifacts, lineage, and experiment reproducibility across parallel runs.
• Work with product, research, and platform teams to connect fine-tuning priorities with business goals.
• Prepare clear documentation for both technical and non-technical audiences.
• Support and coach other engineers on fine-tuning practices, evaluation quality, and responsible deployment.
• Keep up with LLM research and convert new ideas into practical production workflows.

Required Qualifications

• Master’s or PhD in Computer Science, Machine Learning, or a related discipline, or equivalent experience.
• At least 6 years of combined machine learning research and engineering experience, including substantial LLM exposure.
• Strong Python skills and hands-on experience with modern deep learning frameworks, especially PyTorch.
• Experience fine-tuning transformer-based language models at meaningful scale.
• Working knowledge of distributed training methods such as FSDP, ZeRO, and pipeline parallelism.
• Experience with RLHF, DPO, or other preference optimization approaches.
• Solid understanding of evaluation design, benchmarking, and human assessment methods.
• Experience running training workloads on GPU clusters and recovering from system or training failures.
• Strong written and spoken communication skills.
• Evidence of shipping or publishing impactful LLM-related work.

Preferred Qualifications

• Publications in leading machine learning conferences or journals.
• Experience tuning multimodal models.
• Exposure to synthetic data generation and dataset distillation.
• Contributions to open-source LLM training tools or libraries.
• Experience with responsible AI evaluation and red-teaming practices.

Equal Opportunity and Inclusion

Bright Vision Technologies is committed to fair hiring and equal opportunity. The company does not discriminate based on race, religion, color, national origin, gender, sexual orientation, gender identity, gender expression, age, marital status, veteran status, pregnancy, disability, genetic information, or any other protected category under applicable law.

Reasonable accommodations are available for religious practices, beliefs, and mental or physical health needs. The company also prohibits workplace harassment and discrimination.

Additional Information

This position is offered by a no-fee agency. Contact details were provided for direct consideration, and the company referenced its corporate website for more information.