Areas of Interest
Things that I love working on.
Deep learning
Expertise in developing DL frameworks and libraries at various companies used for training, testing, and deployment.
Speech
Speech plays a key role in reaching the next billion users. Acoustic model and meta-models are improving the lives of millions of people.
NLP
Understanding human languages can be tricky for computers. Semantic and phonetic embeddings are the link between text and numbers.
Vision
Vision is the art of seeing invisible. Reasearch and development in Computer Vision is happing at tremendous rate.
ML
Data is the new raw material for the businesses today. Click-prediction and recommendation are the backbone of this industry.
Deploy
AI is useful only when it serves people efficiently. Infra should support deployment on both server and client (mobile) side.
Visualization
Plots and visualization techniques can help in understanding what model is learning and how it is helping in achieving our goals.
Systems
AI is not just about models. It is a cycle of collecting, storing, cleaning, transforming, annotating, learning, testing, debugging, and deploying.
Performance
Users want low latencies and providers want low cost. Knowledge of CPU (SIMD), and GPU (CUDA/OCL) can help achieve that.