Jan 2022 – Oct 2022 · 9 mos

• Working on Venus Adgen- a platform for Ads creators to design the Image Ads automatically. The platform leverages deep learning to generate visually pleasing, highly memorable Ads. This reduces the overall time and effort for manual creation of Ads.
• Responsible for all aspects related to the text component (OCR, Scene text) of Image Ads. Aspects like text detection-recognition, color, saliency, font detection, text prominence, reading order and text named entities.

May 2020 – Dec 2021 · 1 yr 7 mos

• Lead the Edge AI team (13 members) that worked on HS-ODEP - Holosens One stop development and evaluation, algorithm store and one click deployment platform. ( Holosens is Huawei's smart video surveillance camera series)
• Individual contributor to core modules of HS-ODEP (~8k lines of C/C++ code)
• Played a key role in the first version of HS-ODEP to acquire a major customer within 6 months of product initiation.
• Received Huawei's Prestigious Future star award (2021), Two Timely recognition awards (2020, 2021), IPDT Intelligent Vision Award (2021), Best mentor award (2020)
• Design and development of new features (gstreamer based AI pipeline, model pruning, quantisation) for Intelligent Video Surveillance Cameras.
• Apart from individual contribution, involved in requirements gathering, tasks division and assignment, progress tracking and mentoring team members related to Edge AI.
• Point of contact for 38 key HS-ODEP customers for any Edge AI related issues

Jun 2018 – May 2020 · 1 yr 11 mos · Pune Area, India

• Part of team that works on Augmented Reality- Head Up Display and In-Vehicle offline Smart Voice Assistant.
• Neural Network (Object detection, Segmentation and Speech to Text networks) optimization and porting on Qulacomm and Nvidia's Platforms
• Training and fine-tuning of deep learning based Automatic Speech Recognition (Speech to Text) for Smart Voice Assistant.
• Pruning, Quantization of low footprint Neural Networks for edge devices (Jetson, Drive PX2, TI TDA)

Jul 2016 – Jun 2018 · 1 yr 11 mos · Bengaluru, Karnataka, India

• Designing and coding of algorithms related to Public safety and Smart transportation related Video Analytics. (based on both traditional Computer Vision and Deep Learning methods)
• Development/Training of CNN architectures for object classification and localisation using TensorFlow on NVIDIA's high end and embedded GPUs.
• Deployment and Performance analysis of Deep Learning workloads on organisation's HPC platforms.

Jan 2016 – Jun 2016 · 5 mos · Bengaluru, Karnataka, India

• Developed a complete framework for Automation of Stress Test and End User Validation on Intel Media/ Graphics SoC.

May 2012 – Jul 2012 · 2 mos · Greater Hyderabad Area

• Intern at National Remote Sensing Center, Hyderabad (part of ISRO)
• Implemented optical data fusion techniques on satellite images. Methods from Brovey, IHS transform to multi-scale transform fusion methods like wavelet resolution merge.
• Implemented using MATLAB and presented an in-depth comparative study.