I'm a Research Scientist at Rakuten Institute of Technology (Tokyo), with a focus on Robotics. My major fields of interest are Machine Learning and Artificial Intelligence, Robotics and Control Systems; oh and trivia.
- C++, C, Python, Rust, Go, Matlab, Java, CMake
- ROS, PyTorch, Tensorflow, OR-Tools, Kubernetes
- linux, git, bash, Office Tools
- Robotics, Real Time Systems, Distributed Systems, Deep Machine Learning, Reinforcement Learning, Control Systems
Education
- 2012 - 2017
-
Indian Institute of Technology, Bombay (IIT-Bombay)
M.Tech and B.Tech in Aerospace Engineering
- 2007 - 2012
-
Delhi Public School
Senior Secondary and Intermediate High School
Work Experience
02.2020 - Present
02.2020 - Present
Research Scientist
Rakuten Institute of Technology, Tokyo (Tokyo)- R&D for developing IoT and 5G ecosystem
- Designing security and software architecture for fog computing and pushing AI to the edge
- Working on compute offloading for real-time applications
- Developing low computation AI and matching low cost hardware
09.2017 - 02.2020
09.2017 - 02.2020
Research Scientist
Rakuten Institute of Technology, Tokyo (Tokyo)- Working on Autonomous Last Mile Delivery
- Author of Rakuten Robotics Base Toolkit (Rbot) https://github.com/rakutentech/rbot
- Patent applications in progress for 360° camera and localization without GPS
- Worked on integrated SLAM for dynamic indoor and outdoor environments
- Developed prototype of a robot capable of navigating in crowds, roads and cluttered environments along with recognizing traffic signals, using elevators (without API) using an arm and tamper-detection capabilities
- Used AI/ML for object localization, instance segmentation, gesture recognition and extracting text from images
- Assembled and built a 6-man strong team capable of achieving these goals
- Worked on recommending personalized delivery locations optimised to reduce delivery costs
06.2016 - 04.2017
06.2016 - 04.2017
Teaching Assistant
IIT Bombay (Mumbai)- Assisting professors in teaching aerospace design to a batch of 60 fourth-year undergraduate students
- Helping students set mission statement, create preliminary designs and decide milestones for conceptual design on innovative ideas
- Evaluate presentation and technical reports and check for plagiarism
05.2015 - 08.2015
05.2015 - 08.2015
Summer Intern
GE Aviation (Bengaluru)- Simulation and Analysis of Low Cycle Fatigue (LCF)
- Mentors: Mothilal Rengappa and Arvind Rao, GE JFWTC
- Identified regions and conditions for LCF life in the cold frame of commercial aircraft engines
- Automated workflow by parametrisation, macros and scripts to reduce analysis cycle time
- Developed skills in ANSYS Workbench, ICEM, CFX and Siemens NX apart from Proprietary Software
Projects
Static Map Generation with Tracking of Dynamic Obstacles (Master's Project)
- Developed a sensor agnostic probabilistic 3D static map of the environment for mobile robots
- Track and model motion of dynamic objects to facilitate better motion-planning and path-planning
- Map acts as a Distributed, Atomic, Consistent and Isolated source of truth
Efficient Techniques for Static and Dynamic Mapping using Octrees (Bachelor Technical Project)
- Specialized voxel look-up in a probabilistic 3D hierarchical grid map (Octree) sans Hash Tables or Look Up Tables for path-planning and motion-planning
- Achieved 73% faster seek times for neighbouring voxels and 40% less memory usage with a degradation of 8% in Random Node Access Time as compared to the state-of-the-art open source alternative
- Developed heuristic to handle dynamic obstacles in a probabilistic manner
Head of Autonomous Underwater Vehicle Team (AUV-IITB)
- Lead team of 35 students to design and develop a state-of-the-art unmanned AUV capable of localising itself and performing realistic naval missions using visual, inertial, acoustic and depth sensors using thrusters and pneumatics. The project was supported by Defense Research & Development Organization (DRDO)
- RoboSub: Runners up 2016, Semi-Finalists 2012-2015, 2017
- NIOT SAVe: Winner 2017
- Redesigned legacy software stack by developing Time-Constrained Hierarchical State Machine and Generic Control and Data Flow Architecture used for signal and image processing, and drivers
- Managing operations, logistics and knowledge transfer in a 4-tier multi-disciplinary student-run team
- Improving strategy and design via SWOT analysis, Porters four corners model and collaboration with IDC (IIT Bombay) and Naval Research Board (DRDO)
- Spearheading development in Electrical, Mechanical, Public Relations and Software teams
Software Development Team Lead, AUV-IITB
- Designed guidelines and implemented industry standards for testing framework consisting of style checking, static checking, unit testing, integration testing and virtual simulations
- Extended Image Processing framework to incorporate Machine Learning based object localization
- Upgraded Mission Planner using Artificial Ant Colony to dynamically maximize points in the competition
- Supervised and led junior developers for unhindered development of new modules of the software stack like multi-threaded communication framework, localization error resistant 3D map, acoustic signal localization
Full Stack Engineer, AUV-IITB
- Designed and developed modular architecture using Robot Operating System (ROS) to implement Navigator, Mission Planning, Path Planning, Cognition, and 5 DOF PID Control System
- Co-developed drivers for military grade sensors like Doppler Velocity Log (DVL) and Inertial Measurement Unit (IMU) apart from debug interfaces and thread-safe logging library
- Developed a simulator using Gazebo to provide an in-air platform for pre-deployment code-validation