JAIJITESH SURYAPRAKASH
SYSTEMS ARCHITECTURE • AUTONOMOUS SYSTEMS

JAIJITESH SURYAPRAKASH

SYSTEMS • ROBOTICS • COMPUTATIONAL ENGINEERING

PROFILE SPECIFICATION SYSTEMS • AUTONOMOUS ENGINEERING
CORE PROFILE • SYSTEMS ARCHITECTURE

Hello, I'm
Jaijitesh Suryaprakash

Software engineer and systems builder focused on AI agent runtimes, aerodynamics simulation, embedded robotics, and desktop telemetry systems.

DOMAIN AI SYSTEMS • EMBEDDED ROBOTICS
CORE STACK PYTHON, C++, ESP32, FASTAPI
SIMULATION LBM CFD, THREE.JS, WEBGPU
AVAILABILITY OPEN FOR ENGINEERING ROLES
Jaijitesh Suryaprakash
PORTRAIT FRAME AUTHENTIC
EXPERIENCE ENGINEERING
04
Years Engineering Experience
PORTFOLIO SYSTEMS
24+
Projects Constructed
DEPLOYMENTS NODES
12+
Deployed Systems & Nodes
CAPABILITIES SKILLS
30+
Core Engineering Skills

Institutional Trajectory

SECONDARY CERTIFICATION ACCREDITED
FOUNDATIONS • STEM

Higher Secondary Education

Advanced Mathematics • Classical Physics • Computing

Rigorous analytical foundation in differential calculus, vector kinematics, electromagnetism, and algorithmic problem solving.

UNDERGRADUATE DEGREE VIT VELLORE
UNDERGRADUATE • B.TECH

Bachelor of Technology

Vellore Institute of Technology (VIT Vellore)

Information Technology specializing in Autonomous Systems, Distributed Edge Microservices, ROS 2 Kinematics, and Computational Fluid Dynamics.

Production Systems & Kernels

AERODYNAMICS • SIMULATION YEAR 2024 • 2026

WindSim — Aerodynamics Sandbox & CFD Laboratory

Browser-native aerodynamics platform combining reduced-order flight physics (Reynolds drag, dynamic projected area, barometric atmospheric density) with a D3Q19 Lattice Boltzmann Method (LBM)[1]D3Q19 LBM Simulation Framework
Mesoscopic Navier-Stokes Solver • Pressure & Velocity Fields
reference solver, Three.js 3D flow visualization, and WebGPU detection.

CFD SOLVER D3Q19 LBM SOLVER
VISUALIZATION THREE.JS 3D FLOW
PHYSICS MODEL REYNOLDS DRAG • DHO
RUNTIME JAVASCRIPT, WEBGPU
JavaScript Three.js WebGPU LBM D3Q19 IndexedDB Aerodynamics
[ VIEW SOURCE CODE ] →
EMBEDDED ROBOTICS YEAR 2025 • 2026

AutoBot — Autonomous Tracked Robotics Architecture

12-layer embedded robotics architecture on tracked chassis driven by ESP32-WROOM-32[2]12-Layer Embedded Control Stack
Closed-Loop PID with Anti-Windup • Multi-Rate Scheduler 200Hz-1Hz
, featuring quadrature encoders, QMI8658 6-DOF IMU, closed-loop PID velocity regulation with anti-windup, deterministic multi-rate scheduler (200Hz to 1Hz), and live WebSocket telemetry HUD.

CONTROLLER ESP32-WROOM-32
CONTROL LOOP PID ANTI-WINDUP • S-CURVE
SENSORS 6-DOF IMU • ENCODERS
TELEMETRY WEBSOCKET HUD • INA219
C++ ESP32 PlatformIO PID Control WebSockets Robotics HAL
[ VIEW SOURCE CODE ] →
AI AGENT RUNTIME YEAR 2025 • 2026

Berry — Local-First AI Desktop Agent Runtime

Local-first desktop agent connecting LLM reasoning directly to operating system execution[3]Computer-Use Agent Runtime
Chromium Relay Architecture • OS Execution Sandbox
, featuring OS file and terminal automation, Chromium browser relay architecture, vector skill codices, and multi-provider LLM orchestration.

ORCHESTRATION FASTAPI • PYWEBVIEW
AUTOMATION OS COMPUTER-USE • CUA
BROWSER CHROMIUM RELAY
INFERENCE LITELLM • MULTI-PROVIDER
Python FastAPI PyAutoGUI LiteLLM Docker Chromium Relay
[ VIEW SOURCE CODE ] →
REMOTE SENSING • ML YEAR 2025 • 2026

SATQuery-AI — Multimodal Satellite Remote Sensing

Multimodal Earth observation assistant routing satellite queries through a LangGraph orchestrator to Qwen2.5-VL-3B adapted via 4-bit QLoRA, GroundingDINO visual grounding, ChangeFormer V6 bi-temporal change detection, and optical plus SAR fusion.

VLM BACKBONE QWEN2.5-VL 4-BIT QLORA
GROUNDING GROUNDINGDINO-TINY
CHANGE DETECT CHANGEFORMER V6
PIPELINE LANGGRAPH • GEOTIFF CRS
PyTorch LangGraph FastAPI QLoRA ChangeFormer Sentinel-1 • 2
[ VIEW SOURCE CODE ] →
TECHNICAL REFERENCES • FOOTNOTES CITATIONS [1-3]
[1] Discrete Velocity Boltzmann Kinetics (D3Q19 LBM) • Navier-Stokes Incompressible Flow Approximations.
[2] Deterministic Embedded Robotics • 12-Layer Stack • Multi-Rate Task Scheduling (200Hz to 1Hz) • Closed-Loop PID Anti-Windup.
[3] Local-First Agentic Architecture • OS Process Lifecycle Sandbox • Chromium Automation Relay.

Practical Engineering Journey

Changelog and engineering milestones across autonomous robotics, numerical physics, distributed systems, and edge intelligence.

2026

2026

HACKATHON WIN 3RD PLACE OVERALL

ImpactX 3.0 Hackathon & FarmAssist AI

Led the software architecture of a high-concurrency geospatial dispatch and routing engine operating over WebSockets, and constructed an offline edge computer vision agricultural diagnostics pipeline running quantized neural models on a Raspberry Pi embedded controller.

Go WebSockets Edge Vision Raspberry Pi
[ SCHEMATIC FRAME 01 ]
INSERT MEDIA
[ TELEMETRY LOG 02 ]
INSERT MEDIA
[ DISPATCH UI 03 ]
INSERT MEDIA
[ RUNTIME BENCHMARK 04 ]
INSERT MEDIA

2025

2025

EMBEDDED ROBOTICS 12-LAYER ARCHITECTURE

Autonomous Robotics Systems Developer & AutoBot

Engineered a 12-layer embedded robotics architecture on tracked chassis powered by ESP32-WROOM-32, featuring closed-loop PID velocity control with anti-windup, quadrature encoder feedback, QMI8658 6-DOF IMU filtering, multi-rate task scheduling (200Hz to 1Hz), and live WebSocket telemetry streaming.

C++ ESP32 PlatformIO PID Control WebSockets Robotics HAL
[ CHASSIS TELEMETRY 01 ]
INSERT MEDIA
[ PID VELOCITY CONTROL 02 ]
INSERT MEDIA
[ MULTI-RATE SCHEDULER 03 ]
INSERT MEDIA
[ SENSOR BUS IMU 04 ]
INSERT MEDIA

2024

2024

AERODYNAMICS CFD SIMULATION

WindSim & Computational Aerodynamics Developer

Architected browser-native aerodynamics platform featuring reduced-order rigid-body flight dynamics (Reynolds drag, dynamic projected area, barometric atmospheric density model) and a D3Q19 CPU Lattice Boltzmann Method CFD solver paired with interactive Three.js 3D flow visualization and WebGPU detection.

JavaScript Three.js WebGPU LBM D3Q19 IndexedDB Aerodynamics
[ D3Q19 LBM SOLVER 01 ]
INSERT MEDIA
[ REYNOLDS DRAG MODEL 02 ]
INSERT MEDIA
[ THREE.JS FLOW FIELD 03 ]
INSERT MEDIA
[ WEBGPU CAPABILITY 04 ]
INSERT MEDIA

Foundations

Foundations

SYSTEMS • AGENTS CORE ARCHITECTURES

Systems Software, Agent Infrastructure & Applied ML

Developed desktop hardware telemetry suite Performance Monitor (Electron, TypeScript, WMI, Windows performance counters), built from-scratch ReAct agentic RAG engines without external frameworks, tabular synthesis with CTGAN and TVAE, and local desktop agent orchestration in Berry.

Python Electron TypeScript LiteLLM CTGAN Agentic RAG
[ DESKTOP TELEMETRY 01 ]
INSERT MEDIA
[ REACTION LOOP AGENT 02 ]
INSERT MEDIA
[ SYNTHETIC DATA CTGAN 03 ]
INSERT MEDIA
[ LOCAL AGENT RUNTIME 04 ]
INSERT MEDIA
EXPLORATION RESEARCH PERSPECTIVES
ARCHIVE • RESEARCH • SYSTEMS

Do you want to explore the technical research archive?

Inspect technical research notes, reading indices in cybernetics and distributed systems, synthesized lo-fi audio soundscapes, and personal engineering philosophies.

Explore Archive
PORTFOLIO JAIJITESH SURYAPRAKASH
DISCIPLINE SYSTEMS • ROBOTICS
LOCATION VELLORE, INDIA
AVAILABILITY OPEN FOR ROLES