Feb 2011 – Present · 15 yrs 3 mos · India

• Software Engineer & Technical Lead, Data Centre Group (July 2018 – Present)
• Developed and Upstreamed the Intel® Memory Usage Analyzer tool to interface different memory-reclaimers to analyze the workload sensitivity with memory-tiering solutions (https://github.com/intel/memory-usageanalyzer). Leveraged Intel’s Next Gen server accelerator to improve page fault latency of Linux Operating System. Delivered 2x INT8 inference performance vs. FP32 – demonstrating Intel 3rd Generation Xeon Scalable Processors® built-in AI acceleration using DL Boost® key cloud customers such as AWS, Alibaba, Bytedance, etc. on Tensorflow BERT Base and Large XNLI and SQUAD 1.1 models. Lead the team in delivering this kind of first ever Intel optimization on the fp32 graph level performance for TensorFlow Wide-and-Deep Recommender System on Intel 3rd Generation Xeon Scalable Processors® using DL Boost towards an expected $500M market opportunity. Instrumental in achieving 4x FP32 & 2x INT8 performance improvement by optimizing the model. Optimization of the SVT Video codec on the Xeon platform
• Software Engineer, Open-Source Technology Group (Nov 2016 - June 2018)
• Lead the Graphics System Integration team of size 7 for productization of the graphics & display stack and open Sourcing of the Android Graphics stack on Intel Client Platform
• Graphics Software Engineer, Visual and Parallel Computing Group (July 2013 - Oct 2016)
• Designed and implemented robust kernel mode and the user mode components of Android Display Driver for multiple SKUs of Bay trail - BYTT/M/D/CR, Cherrytrail & Broxton platforms and successfully productized these products to customers like ASUS, ACER, CZC, THD, Lenovo.. and worked with Open-source community to upstream those patches.
• Software Engineer, Client Computing Group (Feb 2011 - June 2013)
• Developed Intel Wireless Display Solution WiDi on Android Medfield Tablet platform.