Forward Deployed Engineer, Post-Training

Together AI•San Francisco, CA

About The Position

As a Forward Deployed Engineer (FDE) focused on Post-Training, you will be a hands-on technical partner to strategic customers using fine-tuning and reinforcement learning to build differentiated models on open-source foundations. Working alongside Solutions Architects as a post-training specialist, you will guide customers through data strategy, algorithm selection, experiment design, evaluation, and compute sizing. You will help deliver post-training proofs of concept, drive adoption of the fine-tuning and RL platform, and turn customer goals into measurable gains in model quality. The role directly affects customer success, company growth, and the development of the core platform.

Requirements

At least 5 years in a technical role in ML engineering, Applied AI, Forward Deployed Engineering, or Solutions Architecture, with recent hands-on experience post-training LLMs.
Expert-level, hands-on experience with LLM fine-tuning and/or reinforcement learning, including SFT, LoRA, DPO, and RL methods such as GRPO and PPO.
Strong track record designing experiments, running ablations, and building evaluations that distinguish genuine improvements from noise.
Proven ability to build and iterate on training datasets.
Broad knowledge of state-of-the-art open-source models and strong judgment in selecting a base model for customer requirements.
Strong Python skills and comfort working in production environments.
Experience leading complex customer-facing technical projects, with exceptional project-management and communication skills.
Ability to work in a highly cross-functional environment.

Responsibilities

Help customers define and execute post-training data strategies, diagnose data-quality issues affecting model performance, and recommend approaches for generating data at scale.
Recommend appropriate post-training approaches for customer use cases.
Design controlled experiments and tracking methods; diagnose training instability, reward hacking, and regressions.
Build evaluation frameworks with customers, including task-specific benchmarks, LLM-as-judge pipelines, and held-out test sets, and ensure gains align reproducibly with production success metrics.
Guide base-model selection and baseline quality evaluations before post-training; independently research newly released open-source models, particularly in relation to post-training.
Scope and size RL and fine-tuning workloads, partnering with sales and capacity teams on deals and capacity forecasts.
Act as the primary post-training technical point of contact for aligned strategic accounts; lead technical execution from scoping through production and deliver critical milestones.
Provide field insights to shape the fine-tuning and RL product roadmap.
Contribute code, recipes, and reference implementations as needed, and drive early adoption of new features and research with strategic customers.