I build Robots!• Designed, Developed and Integrated Self-Calibration Tool for Intel Stereo Depth Cameras to Robotics Framework. This helps cameras to detects objects such as Punch, Calibration Fixtures, Punch-Cores, calibrate well and provide optimal performance.• Developed Workflow for computer vision state-model to process images to index Robotic Endeffector and needle. This helps prepare needle for site-making with image processing. Framework uses OpenCV to record and transform images and software stack is developed in C++. • Re-write existing framework RoboCore Windows based subsystems for Linux platform and integrated the service using Kinova Robotics packages on ARTAS IX Robot Controller as Linux Firmware for Robotics Arms Control.• Developing software for Robotics Devices and Control. Some of the features implemented are developing support for data serialization/deserialization for backward compatibility in current ARTAS IX from previous versions. For Data serialization format Boost/C++ framework is used.· Language & Platform: C++, Linux, Windows

Developing self-driving tech• Developed Traffic Sign Labeling framework to process and sensor data from images and storage with meta-data in database engine. C++ Tool and Platform services are developed to help train On-car detection model.• Developed Log Extraction framework for processing different events and critical annotations recorded during the Self- Driving fleet runs. Logs are in raw-formats/video logs and frameworks are developed in C++ with 3D point cloud data and trackLabels for different objects of interest.• Developed and setup VLS framework (Vehicle Lights Sequence) to process Turn Events, Brake States and different Light Sequences for fleets. Framework does the Track Extractions, creates the Job units for Labeling.• Developed IsParked framework to work on logset for Parked Vehicles, Parked but moving, Lane-blocking and similar events and process those annotations. Tools, platforms and services On-Car models to accurately predict behavior of roadside parked vehicles, lane-obstructions and take necessary course in the event.· Language & Platform: C++, Python, Linux

Software Developer in AI/ML, Neural Networks, Deep Learning, Big Data, NLP, Search Engines• Developed a video analytic framework on Computer Vision for Camera Software. For the dataset, Realtime Streaming Video was used with Apache Kafka Cluster, HDFS Filesystem as backend, Apache Spark as Job Cluster, feature extraction Using OpenCV. Project was developed as a Validation Application for Image Processing and Object Detection• Developed a data modeling framework based on Faster RCNN Neural Network that works on data sets, statistics, recommendation and feedback approach. Developed to support many loop runs for training data, applying inference and accuracy rates etc• Developed a recommendation framework for Feeds based Search Service. This involves building framework with designing Distributed WebCrawlers, Index systems, Backend AI - Topic modeling based on Unsupervised Algorithm, Labeling Clusters for visualization and Client interaction Modules• Worked on building mobile search platform project that prepares data for keywords and tokens indexing, storage and serves clients based on query for quick analysis. The model is being developed in 3 versions. In first two versions, indexing and instant autocomplete features are added for query based analysis with TFIDF and Ranking. Currently, working on semantic analysis in third version.Specialties: C/C++/Python, Redis, Multithreading, Wireshark, VDBench, GDB, Apache Solr/Lucene/Kafka, HDFS, TensorFlow/Keras, Py2Neo, Gremlin

• Designed and developed solution for lockfree filesystem for next flash storage version. For lockfree approach, idea was built upon using commit sequence no (CSN) and using lock during commit. This improved overall performance and efficiency by 20% -30 %• Developed solution for managing distributed bufferlists to reduce the lock contention and overall overhead due to central bufferlist. It helped performance jump by 10%-15% in overall architecture• Worked on Buffer Cache redesign and Optimization. Developed code for batch processing of data blocks with memory buckets in hashtable. Each entry was optimized to use and maintain a single linked list of data buffers thus saving of about half of pointers, registers and memory at run time• FileSystems stats logging feature for read/write/readahead IOs and throughput data• Developed statesave feature to save and debug core related info for the failures occurred in case of Panic/HealthCheck/Data Jorunal Cache buffer etc• Developed FileSystem Rehome feature for tackling slow IOs• User End Agent – AntiVirus feature development with Symantec Scan Engine for Oracle Flash StorageFilesystems. Supports CIFS protocol for scan on CIFS Shares for operations like – open/read/write/closeetc· Environment & Platform: C/C++, Linux

Masters' Student in CS at San Jose State University.