Skip to Content

Shaurya Tiwari

AI Software Engineer · Physical AI · OpenUSD · Simulation
Download / View Resume

Summary

AI Software Engineer specializing in production-grade agentic systems and data retrieval platforms working as a forward deployed engineer for 9+ verticals at U.S. News & World Report. Expertise in Django backends, LangChain/LangGraph orchestration, vector store usage (OpenSearch/FAISS), and streaming APIs. Proven ability to ship reliable AI integrated products with industry grade guardrail infrastructure support (legal compliance), from site search to B2B data analysis agentic tool with real time graph and csv generations, performing in ambiguous environments with strong cross-functional collaboration across multiple verticals.

NVIDIA-Certified in OpenUSD, with hands-on Isaac Sim, Isaac Lab, robot policy training, and simulation workflows for Physical AI.

Physical AI & Simulation

NVIDIA-Certified Professional: OpenUSD Development

July 2026

Force-Aware Peg Insertion

Isaac Lab, Isaac Sim, PPO, PhysXGitHub GitHub
  • Benchmarked four PPO policy variants across three seeds to evaluate whether force feedback improves Franka peg-insertion robustness under nominal and randomized conditions.
  • Designed controlled reward/observation ablations with TensorBoard-logged metrics against a reproduced geometry-only baseline.

GPU-Parallel Robot Policy Training

Isaac Lab, Isaac Sim, PhysX, ovphysx, PythonGitHub GitHub
  • Built GPU-parallel reinforcement-learning workflows in Isaac Lab, running large-scale PhysX simulations for robot policy training.
  • Configured vectorized environments, observations, actions, rewards, and domain randomization for high-throughput policy training and evaluation.

Digital-Twin & Robotics Asset Pipeline

Omniverse, OpenUSD, ovrtx, ovphysx, Isaac SimGitHub GitHub
  • Assembled and optimized SimReady OpenUSD digital twins and robotics assets using composition, instancing, materials, URDF interoperability, validation, and scene-optimization workflows.
  • Structured USD environments for modular ovrtx rendering/sensor simulation and ovphysx physics workflows across NVIDIA Physical AI applications.

Experience

U.S. News & World Report

Aug 2024 – Present
AI Software EngineerNew York
  • Architected ByteSage Django backend driving AI products with LangGraph agent workflows
  • Implement RAG pipelines using OpenSearch/FAISS for semantic retrieval with metadata filtering
  • Deploy streaming endpoints; harden reliability across dev/sandbox/UAT/prod environments
  • Build guardrails and conditional routing for production-safe agentic tool execution
  • Partnered with stakeholders across 12 business verticals to convert ambiguous domain problems into validated AI roadmaps, implementation plans, and production-ready engineering scopes

COAPS, Florida State University

May 2023 – May 2024
Software EngineerFlorida
  • Built Python/Airflow pipelines processing 1M+ marine data points into PostgreSQL (10+ TB)
  • Developed Django REST APIs serving 2M+ users/researchers with the marine data acquisition dashboards

Florida State University

Aug 2022 – Apr 2023
Graduate Research AssistantFlorida
  • Engineered PyTorch/TensorFlow models achieving 0.92 F1-score for anomaly detection
  • Reduced inference time 40% via custom CUDA kernels; boosted throughput 30% with GPU clustering
  • Built and trained deep learning models in PyTorch, including GAN-based experiments and reinforcement learning concepts for synthetic climate-state data generation

ACS Pvt. Ltd

Jan 2021 – Dec 2021
DevOps EngineerMumbai
  • Maintained CI/CD pipelines (Jenkins, Docker, SonarQube); cut MTTR 80% with AWS CloudWatch
  • Automated infrastructure provisioning with Terraform, reducing deployment time by 60%

Cart Geek

Jun 2020 – Dec 2020
Web Development InternMumbai
  • It was here where I was introduced to multiple CMS systems like Wordpress, Drupal, Magento, and Shopify.
  • Implemented multiple extension based solutions for client sites supporting increase in traffic and clicks.

Selected Projects

Cerebral Valley Agent Hackathon - Agent IDE Mind Map (Nov 2025)

Using Antigravity IDE’s reasoning logs to generate mind map for tracking the development process with AI agents.

Portfolio AI Chatbot (May 2024)

RAG assistant with Llama 3, Chroma vector DB, LangChain for conversational resume queries.

Transformer from Scratch (Feb–Apr 2024)

Built text completion model implementing self-attention; tracked perplexity dynamics.

Medical Image Segmentation (Aug–Dec 2023)

Polyp detection model (PyTorch/OpenCV) achieving 0.92 DSC, 0.86 JC.

AutoTicket: Internal Company Hackathon (March 2026)

Built an automatic Jira/Confluence/GitLab-integrated triaging system on ByteSage's agent orchestration, generating tickets and boilerplate code from intake forms and saving up to 3 weeks of project initiation time.

POView: Cerebral Valley Hackathon (March 2026)

Built a 3D geospatial globe interface with the Google Places API and Gemini Live API, letting users navigate to referenced locations through conversational AI.

InstruX (Feb 2026 – Present)

External Link instrux.world

Open source project. Designed a reference architecture for an agentic robotics-evaluation layer connecting simulation, synthetic scenario generation, and policy-readiness assessment.

NVIDIA Applied Work

Technical Communication

  • 100+ livestreams hosted on OpenUSD, Isaac Sim, Isaac Lab, and Physical AI concepts, all logged on YouTube Live with code, notes, and references

SIGGRAPH 2026: NVIDIA Physical AI Hands-On Labs

  • Blender, Oranges, and ovrtx: integrated ovrtx, ovphysx, and ovstage with Blender to build an RTX-rendered, physics-enabled OpenUSD workflow using composable NVIDIA Omniverse Libraries
  • Ray Trace Your Way With Omniverse Libraries: built an OpenUSD-based RTX viewport with Omniverse Libraries, loading and querying USD scenes, rendering RTX output, controlling cameras, and generating image results
  • Advancing World Simulation With 3D Gaussian Splatting: built a real-world-to- simulation reconstruction workflow using NVIDIA fVDB and Omniverse NuRec, applying multi-GPU 3D Gaussian Splatting, segmentation, and object-level integration to create simulation-ready environments for Isaac Sim / Isaac Lab

Skills

Languages

  • Python
  • C/C++ (CUDA)
  • Java
  • JavaScript/TypeScript
  • SQL

AI/ML

  • LangChain
  • LangGraph
  • PyTorch
  • TensorFlow
  • RAG
  • Vector Store
  • PPO

Data/Infra

  • Django
  • PostgreSQL
  • Airflow
  • OpenSearch
  • Docker
  • AWS

Tools/Observability

  • Git
  • Jenkins
  • New Relic
  • Snyk
  • CloudWatch

Physical AI / Simulation

  • OpenUSD
  • Isaac Sim
  • Isaac Lab
  • PhysX
  • ovrtx
  • ovphysx
  • ovstage

Education

Florida State University

May 2024
M.S. Computer ScienceGPA: 3.8/4.0

NMIMS University

Aug 2022
B.Tech Information TechnologyGPA: 3.3/4.0

Impact Snapshot

Project / ImpactTech StackCompany & Date
Production AI workflows with tool orchestration and guardrailsDjango, LangChain, LangGraph, DRF APIsU.S. News (Present)
Improved system observability and reduced mean time to detect and report outagesNew Relic, Slackbots, ByteSageU.S. News (Present)
Real-time analytics platform serving 200k+ users across 125 schoolsPython, Airflow, PostgreSQL, Django REST, ReactCOAPS (2023–2024)
High-performance anomaly detection (40% speedup)PyTorch, TensorFlow, CUDA kernelsFSU (2022–2023)
Automated CI/CD smoketests reducing deployment risk 80%Jenkins, Docker, AWS EC2/S3, CloudWatchACS (2021)