click for music
AI Systems Engineer  ·  IIT Jodhpur  ·  CSE

Piyush
Singh Bhati

Building the future at the intersection of
Artificial Intelligence  ×  Computing Hardware  ×  Autonomous Systems

B.Tech CSE · IIT Jodhpur · AI Systems · OS · Robotics
7+
Projects
2
Research
IIT
Jodhpur
scroll
Piyush Singh Bhati
// AI Systems Engineering
Building Infrastructure for ML Workloads

Engineering AI Systems

I am an undergraduate Computer Science student at IIT Jodhpur focused on AI systems, operating systems, and computer architecture.

My work explores how modern AI workloads interact with hardware and system software. I build infrastructure for machine learning — including GPU-accelerated inference pipelines, lightweight operating systems for ML workloads, and distributed AI systems.

My interests include ML systems, GPU architecture, AI inference infrastructure, and operating system design for high-performance and low-power computing environments.

I also have a keen passion for design and aesthetics expressed through CAD Modelling in the domain of Robotics.

AI Systems Operating Systems GPU Architecture Robotics Computer Architecture Embedded Systems

Research Experience

IIT Jodhpur
AI Systems Research
Advisor: Prof. Sidharth Sharma
Working on AI systems and GPU architecture, focusing on system-level optimization for machine learning workloads and studying the NVIDIA Hopper architecture.
AI Systems GPU Architecture ML Optimization NVIDIA Hopper
IIT Jodhpur
Scientific Machine Learning Research
Advisor: Prof. Harshal D. Akolekar
Working on Physics-Informed Neural Networks and CFD surrogate modeling for scientific computing applications.
PINNs Scientific Computing CFD Surrogate Modeling Deep Learning

Projects & Research

ARM64
2025–Ongoing · Embedded Systems / OS

ARM64 Unikernel for ML Inference

Bare-metal ARM64 unikernel (<256KB) for deterministic ML inference with custom ONNX runtime and NEON SIMD kernels achieving ~269µs latency.

View Details
OS Scheduler
2025 · Systems Programming

OS Scheduling Algorithm Simulator

Classical OS scheduling algorithms with interactive UI, real-time Gantt charts, and cross-platform Electron desktop app.

View Details
JARVIS
2025 · AI/ML Project

JARVIS Voice Assistant

Real-time voice assistant with sub-second GPU-accelerated inference on NVIDIA A5000, RAG-based context, and 500+ concurrent user support.

View Details
LLM Charts
2025 · Deep Learning

LLM-Based Chart Generation

Automated chart generation from uploaded PDFs using LLMs and semantic NLP extraction with real-time visualization rendering.

View Details
Crop Detection
2025 · Computer Vision

Crop Stress Detection — U-Net

Pixel-wise semantic segmentation of stressed crops from aerial imagery using a 7.7M-param U-Net with real-time video inference.

View Details
Wing Design
Inter IIT Tech Meet 14.0 · Aeronautics

Fixed-Wing STOL Wing Design

High-lift wing configuration achieving CL = 8.1258 with full CAD assembly and CFD validation using ANSYS Fluent.

View Details
Quadcopter
Personal Project · Robotics

Quadcopter CAD Design

Custom quadcopter with full CAD modeling, structural analysis, and modular component design optimized for autonomous flight.

View Details

Skills & Technologies

Programming
CC++PythonTypeScriptReact
Robotics & Simulation
Multi-rotor DynamicsCFD SimulationsCAD ModelingANSYS FluentElectronics
Machine Learning
Deep LearningANNsCNNs (YOLO, U-Net)Seq-to-Seq Models
Systems
Operating SystemsScheduling AlgorithmsReal-Time Systems
Tools & DevOps
DockerKubernetesGitGitHubVS Code
AI & LLMs
vLLMRAGLangChainGPU InferenceAgentic AI
Academic Foundation — Relevant Coursework
Data Structures & Algorithms Linear Algebra & O.D.E Probability & Stochastic Processes Introduction to Computer Science Software Engineering Pattern Recognition and Machine Learning

Let's Build Something Together

Interested in collaborating on robotics, autonomous systems, or AI projects? I'm always open to discussing new opportunities, research, and ideas.

Ready to Collaborate?

Whether it's a research project, internship opportunity, or just a tech discussion — I'd love to hear from you.

Send Email

Key Achievements