Rohan Rao Email & Phone Number

Mountain View, Ca, Us

Santa Clara, Ca, Us

Tech lead for advanced retrieval and agentic function-calling LLM/RAG workflows, knowledge graphs, multimodal image and video retrieval pipelines. Public domain work on knowledge graphs - NVIDIA dev blog post (https://developer.nvidia.com/blog/insights-techniques-and-evaluation-for-llm-driven-knowledge-graphs), open-source cuGraph-accelerated codebase (https://github.com/NVIDIA/GenerativeAIExamples/tree/main/community/knowledge_graph_rag), GTC Talk by BlackRock (https://www.nvidia.com/en-us/on-demand/session/gtc24-s62356) and co-authored publication on HybridRAG (https://arxiv.org/abs/2408.04948).

Santa Clara, Ca, Us

Building enterprise-grade multimodal AI copilots using retrieval-augmented generation (RAG) for 20+ teams across NVIDIA using a modular framework and internally hosted large language models.Incorporating SOTA models for image, speech, video and document processing, and GPU-accelerated data analytics pipelines. Some of my work in the public domain through NVIDIA:1. Build a Retrieval-Augmented Generation Chatbot in 5 Minutes: https://www.youtube.com/watch?v=N_OOfkEWcOk2. Presented Multimodal RAG Chatbot at GTC 2024: https://www.youtube.com/watch?v=HF4vil1lRV43. Multimodal Assistant Codebase: https://github.com/NVIDIA/GenerativeAIExamples/tree/main/experimental/multimodal_assistant 4. O-RAN Chatbot for Telco: https://github.com/NVIDIA/GenerativeAIExamples/tree/main/experimental/oran-chatbot-multimodal. Featured in NVIDIA Next-Gen Telco Video: https://youtu.be/6H2pOK9QkVk?si=KO3XWMe0bx9HWa9X&t=141.5. CES 2024 in-car assistant: https://www.cerence.com/news-releases/news-release-details/cerence-pioneers-automotive-specific-llm-collaboration-nvidia.6. In-Vehicle Generative AI Powered by NVIDIA Drive: https://www.youtube.com/watch?v=t-UPlPlrYgQ

Santa Clara, Ca, Us

Engineering lead for integration and bring-up of next-gen ATG-B Fleet Software deliveries to Mercedes-Benz and the rollout of ATG-B platforms and software to the entire NVIDIA fleet. This involved cross-functional team collaboration, in-car testing, creating software packages, videos, handbooks and demos for release.

Santa Clara, Ca, Us

Working on full stack autonomous driving systems, spanning perception, planning and simulation and custom NVIDIA DRIVE hardware. First engineer on the team, which grew to 20+ people across various levels within 2 years.Built automated tools and infrastructure for AV software deployment (vehicle setup and testing). Streamlined processes and workflows across cross-functional teams and saved thousands of hours of provisioning effort. These workflows are used by the entire AV org for in-car development and testing.

Pittsburgh, Pennsylvania, Us

Developing systems and techniques for autonomous multi-drone aerial cinematography. Involves working with GPS, mesh networking radios, pose detection, multi-sensor (camera, LIDAR) streaming and tracking, as well as multi-drone/swarm motion planning.Certified commercial drone pilot! (Part 107 Unmanned Aircraft General - Small)

Santa Clara, Ca, Us

Integrated automatic live calibration within the new RoadRunner 2.0 application (based on DriveWorks SDK) for autonomous cars.Designed and implemented a test automation setup using a record-replay-validate gtest-based framework for sensors to ensure self-calibration is successful.Created a custom driving scenario in NVIDIA DRIVE Constellation Simulator and tested it with HIL NVIDIA DRIVE AGX Pegasus across multiple sensor configurations, and was able to perform automated cold-start calibration of all sensors.

Pittsburgh, Pa, Us

Teaching Assistant for a graduate course on Data Mining for MBA students in the Tepper School of Business.

San Francisco, California, Us

Worked on multiple drive by wire enabled vehicles including the Honda Civic 2017 and CRV 2019, BAIC Motors EU4000, and custom robo-vehicle chassis, using combination of open source software and recent research in unstructured environment planning and end to end software+sensor pipelines for constrained driving environments.1. Calibrated multiple LIDARs, GPS and an IMU for creating a mapping vehicle (https://wowelec.wordpress.com/2019/06/18/setting-up-and-calibrating-multiple-lidar-sensors)2. Ran dense point cloud mapping using the Amazon AWS cloud. (https://wowelec.wordpress.com/2019/06/16/running-autoware-based-mapping-in-the-cloud)3. Tested waypoint following with A-star and pure pursuit path planning, and OpenPlanner based dynamic path planning with a hand drawn HD vector map.

Chennai, Tamil Nadu, In

This is my final year master's project as well as something that I have been working on intensely for two years now. It is a fascinating problem to solve, and we have developed some unique solutions that we believe will work in the Indian market.This resulted in work on software defined radio communication with FPGAs, computer vision accelerators on FPGAs, autonomous car perception software development, deep learning optimization for ARM chips with binarization, motion deblurring using generative adversarial networks, semantic segmentation for unstructured Indian road scenes, and algorithms for measuring driver alertness, distraction and productivity.

Santa Clara, Ca, Us

Implemented BFM Flow functionality for the new Xavier System on Chip (SoC) for self-driving cars, to be released in December 2017. Ran various BFM tests on field programmable gate arrays (FPGAs) and wirelessly on networked High Performance ASIC Prototyping Systems (HAPS).Introduced debugging functionality into the silicon and FPGA debugger tool (Tegrashell) for the Vision Processing Unit (VPU) and Programmable Vision Accelerator (PVA) cores in the SoC.

Bangalore, Karnataka, In

Implemented and filed one patent related to the use of machine learning in RF planning and optimization; titled System and Method for Optimal Positioning of Base Station Antenna.Simulated and implemented two patents related to wireless communication circuits that increased spectral efficiency up to 50% with minimal excess hardware.Participated in the Government of India’s Bharat Net project to connect rural areas (nearly 850 million people) to the National Optical Fiber Network (NOFN) as part of the Digital India initiative. Suggested and analyzed modifications in the existing telecommunication standards to adapt to Indian use cases.

New Delhi, Delhi, In

Designed and implemented an air quality monitoring system after integrating various electrochemical sensors and introduced wireless connectivity with WiFi and Bluetooth modules.Was approached by Indian corporates, Blue Star and Godrej, for acquisition of the technology.

Norwalk, Connecticut, Us

Implemented a wireless networked computer vision based scanner application, using Python, the open source OpenCV library, and Google's Tesseract OCR engine.

Master Of Science - Ms, Robotic Systems Development

Dual Degree (B.Tech & M.Tech), Electrical Engineering And Robotics