I am a Firmware Engineer at VIMAAN, building embedded and edge systems for warehouse automation.
I am also a Google Summer of Code 2026 contributor to the Zephyr Project, working on audio infrastructure for the mpipe multimedia framework. My contributions cover I2S and DMIC source elements, capability and buffer negotiation, pipeline lifecycle and xrun recovery, devicetree codec configuration, testing, and hardware validation on the NXP i.MX8MP EVK.
Previously, I worked at Virya Autonomous Technologies on SiMa.ai Modalix perception workloads, GStreamer pipelines, and a human-safety subsystem, and at OLA Electric on embedded AI and ADAS systems across NVIDIA Jetson, TI TDA4VM, and Qualcomm platforms.
I hold a B.Tech in Electronics and Communication Engineering (IoT) from IIIT Nagpur, with a CGPA of 9.13.
- Firmware & RTOS: C, C++, Zephyr RTOS, FreeRTOS, device integration, state machines, structured logging
- Embedded Linux & Build: Embedded Linux, Yocto-based environments, Docker, CMake, West, CI/CD
- Audio & Multimedia: I2S, DMIC/PDM, GStreamer, DeepStream, capability and buffer negotiation
- Robotics & Perception: ROS 2, LiDAR, camera pipelines, edge AI, NVIDIA Jetson, SiMa.ai Modalix
- Platforms & Interfaces: NXP i.MX8MP, TI TDA4VM, Qualcomm QCS5430, CAN, USB, I2C, SPI, UART
As a Google Summer of Code 2026 contributor, I am contributing audio pipeline infrastructure to Zephyr's proposed mpipe multimedia framework.
| Contribution | Link |
|---|---|
| I2S source element, audio capability negotiation, lifecycle handling, and regression coverage | Zephyr PR #114261 |
| Unified DMIC and I2S audio loopback samples with hardware validation | Zephyr PR #114262 |
- GSoC 2026: Zephyr Multimedia Pipeline: Building reusable I2S and DMIC audio-source infrastructure for Zephyr's mpipe framework, including format negotiation, buffer management, lifecycle recovery, and i.MX8MP hardware validation.
- Big Bertha: On-Device RL Quadruped: Team project developing a 12-DOF quadruped with Isaac Lab-trained locomotion, ROS 2 and Nav2, LiDAR SLAM, and on-device ONNX inference on Arduino UNO Q.
- DWA Local Planner: Custom Dynamic Window Approach planner for TurtleBot3 on ROS 2 Humble with obstacle avoidance, multi-goal queuing, and RViz trajectory visualization.