pages: update job description

src/pages/careers.mdx

@@ -56,7 +56,7 @@ import figBrainRoom from './image/brain-room.png';
 
 - Platform-level optimization for large-scale training of Transformer-based models
 - Acceleration/quantization techniques for serving generative models
-- Removing redundant operations for efficient parallel decoding algorithms
+- Parallelization/caching strategies on state-of-the-art decoding algorithms
 
 </TabItem>
 </Tabs>