OPERATOR      : KRISHNA RUSTAGI
DOMAIN        : ROBOTICS & ARTIFICIAL INTELLIGENCE
ACADEMIC BASE : DIT UNIVERSITY, DEHRADUN [B.TECH '24–'28]
PRIMARY FOCUS : HARDWARE-IN-THE-LOOP TESTBEDS • SENSOR FUSION • AGENTIC SYSTEMS
LEADERSHIP    : CO-FOUNDER & VICE PRESIDENT OF AI @ ASTRA ROBOTICS & AI CLUB
STACK PROFILE : ESP32 • ROS • DUAL I2C • C/C++ • PYTHON • REACT 18 • FASTAPI • CAD
SYSTEM STATUS : ACTIVE // BUILDING PHYSICAL EMBEDDED & MULTI-AGENT PIPELINES
ACCREDITATION : VERIFIED IBM, ANTHROPIC, SOLIDWORKS (30H CAD), VERGEGEN ROS
LOCATION      : DEHRADUN, UTTARAKHAND, INDIA [30.3165° N, 78.0322° E]

I am a Robotics & Artificial Intelligence undergraduate at DIT University engineering real-world mechatronics, multi-sensor telemetry pipelines, and agentic machine learning systems.

• What I Build: End-to-end hardware-in-the-loop diagnostic testbeds, dual-bus microcontroller firmware, multi-class sensor fusion classification models, and deterministic DAG agent workflows.
• Engineering Philosophy: Hardware-software co-design. I bridge low-level micro-architecture (registers, PWM frequencies, I2C bus arbitration, high-speed sampling) with high-level software abstractions (FastAPI WebSockets, scikit-learn consensus inference, React state graphs).
• Current Focus: Real-time sensor fusion algorithms, high-frequency IMU vibration telemetry, and agentic reasoning architectures for robotics control.

┌──────────────────────────────────────────────────────────────────────────────────────────┐
│ 01 // BLDC MOTOR AI PREDICTIVE MAINTENANCE RIG                                           │
└──────────────────────────────────────────────────────────────────────────────────────────┘

Real-time hardware-in-the-loop motor diagnostic rig streaming 6-sensor telemetry over dual independent I2C buses with consensus machine learning fault detection.

• Tech Stack: C/C++ (ESP32 Dev Module) • Python • FastAPI • WebSockets • Dual I2C • Scikit-Learn • SQLite
• Engineering Depth:
  • Designed a dual hardware I2C bus configuration (Bus 0 on GPIO 21/22 and Bus 1 on GPIO 17/16), completely eliminating address collisions for dual GM009605 OLED displays (both address 0x3C) without external multiplexers.
  • Interfaced 6 onboard sensors: ISM330DHCX 6-DoF IMU, MMC5983MA 3-axis magnetometer, INA219 high-side current/voltage, LM35 analog core temperature, optical IR tachometer, and 30A ESC.
  • Implemented a Diagnostic Consensus Engine in fusion.py fusing Decision Tree (85–95% accuracy), Logistic Regression, and K-Means with deterministic physical rule overrides and an anomaly detector (thermal shock, INA219 dropouts).
  • Built non-blocking firmware sampling at 20Hz and delivered telemetry across three surfaces: onboard OLEDs, local desktop GUI (Tkinter), and public FastAPI/WebSocket cloud dashboard.
• Status: OPERATIONAL // HARDWARE DEPLOYED
• Links: Source Code & Architecture • Full Wiring Guide

┌──────────────────────────────────────────────────────────────────────────────────────────┐
│ 02 // HEXFLOW STUDIO — AGENTIC NODE DIRECTOR & CONTINUITY CANVAS                         │
└──────────────────────────────────────────────────────────────────────────────────────────┘

Modular 7-node Directed Acyclic Graph (DAG) cinematography engine with LoRA cross-attention identity preservation and an Agentic DP copilot.

• Tech Stack: React 18.3 • Vite 5.4 • DAG Node Canvas • LoRA Vectors • IP-Adapter • Vanilla CSS Design System
• Engineering Depth:
  • Architected a 7-node DAG orchestration engine: Script & Scene, Character Anchor, Camera Optics, Lighting & Film Grade, Motion Dynamics, Magnific Detail Pass, and Master Render 4K.
  • Implemented cross-attention vector binding with LoRA and IP-Adapter weights, achieving 98.4% facial and wardrobe identity consistency across multi-shot transitions ($<0.02$ latent drift).
  • Developed an Agentic Director of Photography (DP) reasoning copilot enforcing optical spatial constraints (180° eyeline rule, 4:1 chiaroscuro contrast ratios, anamorphic bokeh depth).
  • Engineered a high-performance client runtime with custom cubic Bezier cable rendering, sub-second node mutation propagation, and 24 FPS CinemaScope timeline playback.
• Status: PRODUCTION PROTOTYPE // DEPLOYED
• Links: Source Code

┌──────────────────────────────────────────────────────────────────────────────────────────┐
│ 03 // KM-WÄCHTER — FLEET DIAGNOSTIC CODE MODERNIZATION (IBM BOB CHALLENGE)              │
└──────────────────────────────────────────────────────────────────────────────────────────┘

Refactored 2013-era procedural fleet maintenance service for 6,000 vehicles, achieving 100% test automation and predictive breakdown modeling.

• Tech Stack: Python • Pytest • Pandas • Data Pipeline Validation • Fleet Telemetry
• Engineering Depth:
  • Diagnosed and resolved 6 critical legacy software bugs, including inverted km_to_miles constant calculations, integer wear flooring, missing sensor telemetry crashes, and zero-division errors in nightly batch reports.
  • Shifted the test suite from failing to 100% pass rate (4/4 Pytest suites passing; 11/11 mechanical verification gates in verify.py).
  • Performed empirical data analysis on 6,000+ vehicle service records in Pandas (analyze.py). Proved that vehicle age and total mileage have near-zero correlation with breakdowns; isolated km_since_service (+61%), avg_daily_km (+22%), and load_factor (+19%) as the true breakdown drivers.
  • Packaged a local telemetry monitoring web dashboard (index.html) for real-time fleet health visualization.
• Status: COMPLETED & VERIFIED // CREDENTIAL ID: PRV-2026-47775604
• Links: Source Code & Defect Autopsy

┌──────────────────────────────────────────────────────────────────────────────────────────┐
│ 04 // AETHER AI — GENAI SAAS DASHBOARD & MODEL PLAYGROUND                                │
└──────────────────────────────────────────────────────────────────────────────────────────┘

High-performance developer dashboard featuring Google Gemini API integration, dynamic audio synthesis, and credit telemetry.

• Tech Stack: JavaScript (ES6+) • Google Gemini REST API • Web Audio API • HTML5/CSS3 Glassmorphism
• Engineering Depth:
  • Built an interactive model playground with live parameter adjustment for temperature, top-k sampling, and configurable persona tones.
  • Engineered a client-side speech synthesis audio engine with dynamic real-time canvas waveform visualizations and voice profile controls.
  • Implemented client-side micro-SaaS logic validator and real-time token expenditure / credit accounting telemetry.
• Status: FUNCTIONAL ENGINE
• Links: Source Code

┌──────────────────────────────────────────────────────────────────────────────────────────┐
│ 05 // AI COMPUTER VISION DETECTION & AUTOMATION MODEL                                    │
└──────────────────────────────────────────────────────────────────────────────────────────┘

Real-time object localization and automated spatial decision triggering pipeline.

• Tech Stack: Python • OpenCV • Computer Vision Inference • Dynamic Thresholding
• Engineering Depth:
  • Developed an automated computer vision inference pipeline for spatial localization and automated decision triggering.
  • Implemented adaptive image pre-processing, dynamic thresholding, and contour bounding-box extraction resilient across volatile lighting conditions.
  • Profiled frame inference bottlenecks and reduced end-to-end processing latency by 25%.
• Status: BENCHMARK TESTED

┌──────────────────────────────────────────────────────────────────────────────────────────┐
│ 06 // MECHANICAL CAD ASSEMBLY & KINEMATICS PROTOTYPING                                   │
└──────────────────────────────────────────────────────────────────────────────────────────┘

Certified 30-hour precision 3D mechanical modeling, kinematic linkages, and rapid fabrication enclosures.

• Tech Stack: SOLIDWORKS (30-Hours Certified) • Kinematic Analysis • Tolerance Stack-up • DFM
• Engineering Depth:
  • Modeled precision 3D mechanical linkages, motor mounting brackets, sensor enclosures, and structural chassis components.
  • Executed tolerance stack-up analysis and kinematic clearance validation for physical rapid prototyping and assembly.
• Status: CERTIFIED CAD TRAINING COMPLETED

Below is the verified hardware-to-cloud telemetry flow engineered for the BLDC Predictive Maintenance testbed:

[BLDC Motor + 30A ESC] ──(Vibration / Mag / Current / Temp / RPM)
                                  │
                                  ▼
                ┌───────────────────────────────────┐
                │        ESP32 WROOM-32             │
                │  • Bus 0: IMU, Mag, INA219, OLED1 │
                │  • Bus 1: OLED2 (0x3C Collision)  │
                │  • GPIO34: LM35  |  GPIO27: IR    │
                └─────────────────┬─────────────────┘
                                  │ Serialized 20Hz Telemetry
                                  ▼
                ┌───────────────────────────────────┐
                │   DIAGNOSTIC CONSENSUS ENGINE     │
                │  • Decision Tree (85-95% Acc)     │
                │  • Logistic Regression Classifier │
                │  • K-Means Unsupervised Cluster   │
                │  • Physical Safety Rule Override  │
                └─────────────────┬─────────────────┘
                                  │
         ┌────────────────────────┼────────────────────────┐
         ▼                        ▼                        ▼
┌──────────────────┐    ┌──────────────────┐    ┌────────────────────┐
│ Onboard Dual     │    │ Local Desktop    │    │ FastAPI Cloud      │
│ OLED Displays    │    │ GUI (Tkinter)    │    │ WebSocket Stream   │
└──────────────────┘    └──────────────────┘    └────────────────────┘

ROBOTICS & MECHATRONICS
├── Robot Operating System (ROS) [VergeGen Certified]
├── Kinematics, Dynamics & Actuator Control
├── Brushless DC (BLDC) Motors & 30A ESC Calibration
├── Multi-Sensor Telemetry & Hardware-in-the-Loop Diagnostic Rigs
└── Proteus Simulation & Circuit Modeling

HARDWARE & EMBEDDED SYSTEMS
├── Microcontrollers: ESP32 (WROOM-32), Arduino
├── Bus Architecture: Dual Independent I2C (Bus 0 & Bus 1), SPI, UART, PWM
├── Sensors: ISM330DHCX (6-DoF IMU), MMC5983MA (3-Axis Mag), INA219 (Current/Volt)
├── Transducers: LM35 Analog Core Temp, Optical IR High-Speed Tachometer
└── Display Interfaces: Dual SSD1306 / GM009605 OLED Panels

PROGRAMMING & FIRMWARE
├── C / C++ (Microcontroller Firmware, Non-Blocking Polling Loops)
├── Python 3 (FastAPI, Scikit-Learn, Pandas, NumPy, Pytest)
├── JavaScript / TypeScript (ES6+, React 18, Vite)
└── Shell / Linux / Git Version Control

AI, MACHINE LEARNING & AGENT ARCHITECTURES
├── Classification: Decision Trees, Logistic Regression, K-Means Clustering
├── Diagnostics: Consensus Voting Engines, Physical Anomaly Detection
├── Agentic AI: Directed Acyclic Graph (DAG) Workflows, Anthropic Agent Skills
└── Diffusion Opts: LoRA Embeddings, IP-Adapter Vector Cross-Attention

CAD, FABRICATION & TESTING
├── SOLIDWORKS (30-Hours Certified Mechanical Design & Kinematics)
├── Tolerance Stack-up & Physical Clearance Verification
├── Automated Software Testing (Pytest Suite Architecture, 100% Pass Rates)
└── UI/UX Architecture & Schematics (Figma, Modern CSS Systems)

• ROS (Robot Operating System) Training Programme — VergeGen Tech Private Limited (Skills: ROS, Proteus Simulation, Arduino)
• SOLIDWORKS Mechanical CAD Certification — 30 Hours Certified Mechanical Design, Part Modeling, Assembly & Kinematics
• AI-Assisted Code Modernization — IBM & ProoV by Projectstudy.in (Credential ID: PRV-2026-47775604)
• Anthropic AI Certification: Introduction to Agent Skills — Verified Anthropic Credential (Credential ID: vxuzs4wj9ycf)
• Anthropic AI Certification: AI Capabilities and Limitations — Verified Anthropic Credential (Credential ID: 7pua5pz4ob9i)
• Anthropic AI Certification: Introduction to Claude Cowork & Claude 101 — Anthropic Training Completion
• Hack This Fall 2024 — Virtual Hackathon Participant (Rapid Prototyping & System Design)
• Accenture Nordics Job Simulation — Certified Consultant Simulation (Software Architecture & Testing)

• ASTRA x DIT — Robotics & AI Innovation Club, DIT University
  • Co-Founder & Vice President of Artificial Intelligence [Jan 2026 – Present]
  • Spearheaded 10+ hands-on technical workshops, hackathons, and hardware hack days.
  • Mentored undergraduate cohorts in embedded microcontroller programming (ESP32/Arduino), sensor integration, and rapid prototyping. Expanded cross-departmental technical engagement by 40%.
• Microsoft Student Community x DIT
  • Design Head & Technical Team Member [Aug 2024 – Present]
  • Directed design architecture and technical coordination across a 200+ member chapter; accelerated project delivery turnaround by 30%.
• Under25 / DIT University
  • Festival Director & Under25 Fellow [Nov 2025 – Present]
  • Directed operations, technical setups, and cross-functional teams under tight timeline constraints.
• VELNORA
  • Creative Director [Aug 2024 – Present]
  • Supervised brand identity, design layouts, and web interfaces with Figma and modular CSS tokens.

┌── [TRANSMISSION_CHANNELS] ────────────────────────────────────────────────────────┐
│                                                                                   │
│  EMAIL       : krishnarustagi229@gmail.com                                        │
│  LINKEDIN    : https://linkedin.com/in/krishna-rustagi-a83084308                  │
│  GITHUB      : https://github.com/krishnarustagi68-cell                           │
│  ACADEMIC    : DIT University, Mussoorie-Diversion Road, Dehradun, India           │
│                                                                                   │
└───────────────────────────────────────────────────────────────────────────────────┘