Jan 2020 – Dec 2021 · 1 yr 11 mos

• Works as a Computer vision engineer for Advanced Driver Assistance System (ADAS) domain.
• Worked on Computer vision projects which includes image classification, CNN, Object detection, gesture detection.
• Expertise in CUDA, OpenCV, OpenCL programming for GPU acceleration for Automotive Vehicle Projects.

Jan 2017 – Dec 2019 · 2 yrs 11 mos

• Member of the core Android Mobile Camera SW team.
• Served as a key member in SW team for Mobile Camera in Premium and Mid & value tiers for Qualcomm Snapdragon chipsets.
• Worked as part of Design meeting for low-end chipset.
• Worked in User Mode for the Mobile Camera subsystem which is redesigned after HAL1 & HAL3.
• Worked in Sensor and ISP drivers, Android Upgrade like Android O, Android P, Android Q.
• Worked on Single Camera, Dual Camera & Multi Camera features.
• Delivered the QUADCFA sensor features, multiframe feature, High Dynamic Range (HDR, SHDR) to customer.
• Worked on key features for VIVO, OPPO, Oneplus, Samsung, LG phones that included Super Slow motion recording, HFR, HDR, MFNR, QUADCFA.
• Solved complex issues which included ‘Race Condition’, 'Dead lock', 'Memory corruption'
• Solved stability issues which involved analysis on the issue with just one call stack.
• Works as a Camera Subsystem Sensor, serializer, deserializer for Advanced Driver Assistance System (ADAS) domain.
• Worked on safety hardware drivers for ADAS Domain using QNX.

Aug 2014 – Aug 2016 · 2 yrs · Mangalore Area, India

Aug 2014 – Aug 2015 · 1 yr · India

• Environment :Matlab and Java
• Description : To estimate the plaque burden (total plaque area) in the bulb and sinus of the carotid artery. Further, plaque thickness will also be calculated automatically given the contours of the plaque thickness region.Automated segmentation tool will be developed in the region of interest of the carotid ultrasound bifurcation images.

Jan 2013 – Apr 2013 · 3 mos · Kolkota

• Advisor : Dr. Ashish Ghosh, Professor, Machine Intelligence Unit
• Indian Statistical Institute Kolkota
• Environment : MATLAB and JAVA
• Description : Background subtraction is a widely used method in computer vision for separating or segmenting out the foreground objects from the background of a video. The foreground objects are defined to be the parts of the image that changes and the background is made out of the pixels that stay relatively constant.The background subtraction involves two different tasks, each of which needs to be performed real-time with having only the video frames as the input.
• Learning the background model
• Classifying pixels as background or foreground
• These are the algorithm techniques for Background Subtraction which we used in our project
• Frame Differencing
• Running Average Filter - Background Median Filter
• Gaussian Mixture Models (GMM)(for multiple Background)
• Change detection for moving object segmentation with robust background construction under wronskian framework.

Jan 2013 – Apr 2013 · 3 mos · Kolkata, West Bengal, India