Syftr is an open-source framework designed to optimize generative AI workflows by automatically identifying Pareto-optimal configurations that balance accuracy, cost, and latency. Utilizing multi-objective Bayesian Optimization, syftr allows AI teams to efficiently explore workflow options, significantly reducing the complexity and computational cost of evaluating numerous configurations. The framework supports modular customization and integrates with various open-source libraries to enhance AI workflow design.

GEMCODE is a novel pipeline that combines deep generative models and evolutionary optimization for the automated design of co-crystals, aimed at enhancing tabletability for pharmaceutical applications. The system demonstrates effectiveness in exploring chemical spaces under realistic constraints and has successfully predicted new co-crystals that could accelerate drug development. Experimental studies validate its capabilities and potential use of language models in co-crystal generation is also discussed.

PyTorch and vLLM have been integrated to enhance generative AI applications by implementing Prefill/Decode Disaggregation, which improves inference efficiency at scale. This collaboration has optimized Meta's internal inference stack by allowing independent scaling of prefill and decode processes, resulting in better performance metrics. Key optimizations include enhanced KV cache transfer and load balancing, ultimately leading to reduced latency and increased throughput.