Mind Lab opens its Mint Recursive post-training platform to enterprises, with forward-deployed engineers available to run projects end to end

Mind Lab, a San Francisco research lab focused on continual learning and post-training, announced on October 21 that its Mint Recursive platform is now open to enterprises that want AI models trained on their own industry expertise, in sectors such as pharma, finance, the public sector and gaming. Each engagement starts by building a benchmark with the customer that defines the task, what correct behavior looks like, common mistakes and the bar a model must clear; data work, post-training and deployment then run in one loop measured against it.

The company offers three ways to work. Its forward-deployed engineers (FDEs) can run the benchmark, data, post-training and deployment end to end. Teams that define their own model, data and configuration can hand training and delivery to the platform's automation, and teams wanting full control can write their own loss functions and training loops on Mind Lab's infrastructure. Mind Lab says specialist training starts at $1,000 and that GPU costs on shipped projects have fallen by more than 90%, using LoRA and open base models such as GLM, DeepSeek, Kimi, Qwen and MiniMax.

Why it matters for FDEs: post-training is becoming another place where FDE work shows up, following Together AI's similar move earlier this week. Here the FDE's job is less about wiring up integrations and more about turning a customer's expert judgment into benchmarks, training data and evaluation criteria, a skill set that blends ML engineering with domain discovery.