BENGALURU, INDIAOPEN TO WORK ↗

AYUSHKUMAR
JHA.

THE HUMAN BEHIND THE CODE

ECE STUDENT & BUILDER
SCROLL TO EXPLORE ↓

A PORTFOLIO IN CONSTANT EVOLUTION / 2026

01 / A LITTLE CONTEXT

Curiosity meets
working code.

I’m Ayush, an Electronics & Communication Engineering student at SMVIT, Bengaluru. I build across web development, cybersecurity, computer vision, and embedded systems.

READ MY CV ↗

THE TOOLS BEHIND THE WORK

01 / WEB DEVELOPMENT

Daily Routine
Tracker.

JavaScript · IndexedDB · HTML / CSS

DAILY ROUTINE TRACKER ↘

02 / CYBERSECURITY

Cybersecurity
Threat Lab.

Nmap · Wireshark · Kali Linux

CYBERSECURITY THREAT LAB ↘

03 / EMBEDDED SYSTEMS

ESP32
Smart Car.

C++ · ESP32 · Arduino

ESP32 SMART CAR ↘

04 / COMPUTER VISION

Hand Gesture
Recognition.

Python · OpenCV · MediaPipe

HAND GESTURE RECOGNITION ↘

KEEP SCROLLING ↓

02 / EXPERIMENTS MADE REAL

Selected work.⁽⁰⁴⁾

DAILY / JOURNAL MONDAY

A little better.
Every day.

80DAILY SCORE

✓ Move your body ✓ Read & reflect

INTERFACE CONCEPT / WEB DEVELOPMENT

01 / WEB APPLICATION

Daily Routine
Tracker.

A personal workspace for daily intentions, progress charts, and portable backups. Your records stay on your device.

DEVICE-LOCALYour data, with you.PORTABLECSV & JSON exports.

JAVASCRIPT · INDEXEDDB · HTML / CSS

LIVE WEBSITE ↗SOURCE ↗

THE STORY BEHIND THE BUILD

Finding a rhythm that stuck.

THE START

A friend and I were struggling to stay disciplined with our work. We tried notes apps, habit trackers, and Excel spreadsheets, but none felt like a routine we could keep returning to.

THE BUILD

That frustration became the Daily Routine Tracker: one place for daily intentions, progress charts, and reflection. The current version keeps records on the device and supports portable backups.

THE IDEA

The aim is simple: make the next small action easier to see, and make progress easier to revisit.

THREAT LAB / RESEARCH

01 RECONNAISSANCE
02 ASSESSMENT
03 DETECTION
04 MITIGATION

FOUR VOLUMES / ONE SECURITY MINDSET

02 / CYBERSECURITY

Cybersecurity
Threat Lab.

Network reconnaissance, OWASP assessments, and simulated SOC workflows, documented in a four-volume audit series.

RESEARCHFour audit volumes.METHODAssess. Document. Mitigate.

NMAP · WIRESHARK · KALI LINUX

EXPLORE REPOSITORY ↗

THE STORY BEHIND THE BUILD

From course material to investigation.

THE START

The Cybersecurity Threat Lab began as a project for my cybersecurity course.

THE WORK

It gave the tools a practical context: reconnaissance, security assessments, simulated SOC workflows, and written findings. The repository brings that work together as a four-volume audit series.

THE TAKEAWAY

The project connects technical observations with clear documentation and mitigation recommendations.

ESP
32BLUETOOTH → PWM → MOTION

SENSE / DECIDE / ACT

03 / EMBEDDED SYSTEMS

Small car.
Real control.

An Android-controlled ESP32 smart car with ultrasonic obstacle avoidance and PWM motor control.

INPUTBluetooth commands.RESPONSESensor-based control.

C++ · ESP32 · ARDUINO · BLUETOOTH

EXPLORE REPOSITORY ↗

THE STORY BEHIND THE BUILD

One car. Several ways to control it.

OPEN DEMO FOLDER ↗

THE START

The brief was to build an Android-controlled wireless car during a five-day college Skill Lab.

THE EXTENSIONS

Starting with Bluetooth control, I explored mobile voice commands and obstacle avoidance using an ultrasonic sensor on a rotating mechanism. The repository also includes distance measurement and range-scanning work.

THE TAKEAWAY

This was a hands-on introduction to connecting software commands with sensors, motors, and real-world movement.

21 LANDMARKS / REAL-TIME RECOGNITION

04 / COMPUTER VISION

Human gestures.
Machine vision.

Turning webcam input into finger counts with hand-landmark detection and custom finger-state classification.

INPUTLive webcam frames.METHOD21 hand landmarks.

PYTHON · OPENCV · MEDIAPIPE

EXPLORE REPOSITORY ↗

THE STORY BEHIND THE BUILD

A first step into computer vision.

OPEN DEMO FOLDER ↗

THE START

I built this small project while learning OpenCV in my first year.

THE BUILD

Webcam frames become hand landmarks, and finger-state classification turns those landmarks into a visible count.

THE TAKEAWAY

A focused learning experiment: understand the input, inspect the landmarks, and turn them into a useful response.

CURRENTLY BUILDING / WITH A FRIEND

AgentPGO.

Understand an agent.
Then improve how it runs.

We’re working on profile-guided optimization for AI agents: collecting execution metadata, evaluating behaviour, and producing optimization recommendations.

The current backend includes a FastAPI service, tracing connectors, Python and TypeScript SDKs, and evaluation pipelines. It produces recommendations and YAML exports; it is still a work in progress.

FOLLOW THE BUILD ON GITHUB ↗

01 / TRACE→02 / PROFILE→03 / EVALUATE→04 / RECOMMEND

03 / STILL LEARNING. ALWAYS BUILDING.

The next
chapter.

2024 — 2028 EXPECTED: B.E. Electronics & Communication
SMVIT / VTU · CGPA 8.8 / 10
SPECIALISATION: Cybersecurity minor
IIT Guwahati
COMMUNITY: TechHub, SMVIT
Technical events & community coordination

DOWNLOAD FULL CV ↓

BEYOND THE SCREEN