Senior AI Engineer, Post-Training
About this role
Engineers who thrive on turning complex machine learning research into practical production systems will find a strong fit within this legal tech environment. Working inside Carta Law, the team builds autonomous AI agents, specialized legal models, and automated contract workflows designed to streamline private capital administration. This person will take ownership of the entire model development lifecycle, from post-training open-weight language models on proprietary data to deploying inference systems and agent frameworks.
The role exists to drive model-centric projects that enhance document intelligence and legal operations while collaborating closely with domain experts and product engineers. Ideal candidates bring deep technical proficiency in large language model post-training, strong architectural judgment, and a track record of shipping applied AI solutions to production environments.
What you will do
- Post-train open-weight language models on proprietary legal data by managing the end-to-end lifecycle from base model selection to iterative evaluation.
- Apply targeted training methods such as supervised fine-tuning, preference optimization, and reinforcement learning while refining reward mechanisms.
- Develop and maintain training datasets and automated pipelines that incorporate human feedback loops and labeling guidance.
- Manage the training infrastructure for reliable experiment execution and diagnose distributed training jobs to optimize performance.
- Deploy and operate production models, agent systems, evaluation pipelines, and associated software tooling.
- Collaborate with product engineers on system co-design to determine the division of labor between model weights, agent harnesses, and workflows.
- Consult directly with lawyers and domain experts to translate operational legal processes into technical data and evaluation requirements.
What you need
- Possess hands-on experience with large language model post-training using frameworks like PyTorch alongside associated training, evaluation, and inference stacks.
- Demonstrate the ability to construct the surrounding product infrastructure, including agents, tools, and production microservices.
- Maintain up-to-date knowledge of open-weight models and advanced post-training methodologies.
- Show a history of owning model development initiatives in applied settings and shipping functional AI systems to users.
- Exhibit strong technical judgment regarding model selection, data pipelines, training objectives, and trade-offs between model training and agent improvements.
Benefits
- Market competitive salary
- Equity for all full-time roles
- Exceptional benefits
Skills and tech stack
Questions candidates ask
What is the salary range for this position?
The minimum cash compensation range is $242,250 to $285,000 in San Francisco, CA and New York City, New York. Final offers may vary depending on geography, candidate experience, expertise, and other factors.
What team will I be working with?
You will join Carta’s ML Engineering team, which is embedded in Carta Law. This legal tech platform builds autonomous AI agents, specialized legal models, document intelligence, and contract workflows.
What tools and frameworks are required for the role?
The role requires hands-on experience with LLM post-training using PyTorch or equivalent frameworks. You will also work with managed or self-hosted training infrastructure, model serving systems, and agent tooling.