About Stochastic
Vision
Stochastic's vision is to build a personal AI for everyone on the planet. Just as computers evolved from centralized, enterprise-only form factors to personal computers, we believe in the future of personal AI that can help everyone in their day-to-day work.
We believe that the current approach of scaling language models infinitely larger taken by the closed AGI companies causes centralization of AI power, does not meet with human's desire for privacy and individuality, and further accelerates carbon emission problems.
By focusing on smaller, more efficient language models that learn on your knowledge, context, and your interactions, we are planning to deliver useful AI that can run across all devices and be personalized for you and your work.
Team
Founded by Harvard University AI systems researchers that built the world's first Bayesian and LLM inference chips with the full software stack including models and compiler on top, Stochastic is joined by AI researchers and engineers with Masters and PhDs with the passion for making AI more accessible to everyone. Our recent research includes latency-optimized transformers architecture, quantized parameter efficient fine-tuning, and sparsity-aware throughput maximization on GPUs.