Portrait of Vikrant Sharma

Data Engineering and Machine Learning

I build data systems that learn

Vikrant Sharma. Data engineer and machine learning builder, specialised in cybersecurity.

scroll down

Who I am

I move data, clean it and make it think

I am Vikrant Sharma, a Master of IT student at the University of the Sunshine Coast in Adelaide. I build data pipelines and the machine learning models that sit on top of them: pulling raw data in, shaping it and turning it into answers people can act on. On the side I point the same skills at networks and security, where my hybrid detection model scores 0.95 F1 on real attack traffic. Everything I build ships as working software you can click on, not slides.

see the work
0parsers feeding my personal data pipeline
0live model demos online
0tests green on my tracker
0products used daily by real people
AWS Certified Data Engineer Associate Oracle Data Science Professional NASA Space Apps Global Finalist 2023

Featured projects

Raw data in. Working software out.

money.vikrant69g.com
Money tracker dashboard with monthly in and out charts
01

Personal finance data pipeline

Reads my bank alerts from Gmail, runs them through 16 format parsers and lands them in a private dashboard. Encrypted end to end, so the server only ever stores scrambled text. 65 tests green.

  • TypeScript
  • Cloudflare Workers
  • Cryptography
see it live
vikrant892-threat-lens.hf.space
ThreatLens threat intelligence dashboard with live CVE feed
02

ThreatLens intelligence feed

My side interest in security, taken seriously. Streams vulnerability data from the national database, scores it, maps it to attack techniques and plots where attacks come from.

  • Python
  • Data feeds
  • Dashboards
open live demo
vikrant892-nasa-space-challenge.hf.space
Cosmic Keys app turning planetary data into piano music
03

Cosmic Keys

The project behind my NASA Space Apps global finalist run. Takes real planetary data and turns it into piano music: pick planets, hear their numbers become notes.

  • Python
  • Sonification
  • Streamlit
open live demo
Intrusion detection dashboard

Hybrid intrusion detection

Three models spotting hostile network traffic. 0.95 F1, 0.98 ROC AUC on CICIDS2017.

live, password gated open
Phishing detection platform

Phishing detection

Flags dodgy links and lookalike sites before you click them.

moving to Cloudflare open
Credit card fraud detection app

Card fraud detection

Learns spending patterns and calls out transactions that do not fit.

moving to Cloudflare open
Password strength analyser

Password strength API

Scores passwords the way an attacker would, not with a naive checklist.

moving to Cloudflare open
Log analyzer dashboard

Log analyzer

Chews through server logs and surfaces the events worth reading.

moving to Cloudflare open
Job hunter app lock screen

Job application machine

Finds Adelaide internships, writes outreach, queues every email behind human approval. 45 tests green.

private tool GitHub

Hosted on Hugging Face. A sleeping demo takes about 30 seconds to wake up. The ones marked "moving to Cloudflare" hit the free CPU limit and are being rebuilt as always-on Cloudflare apps.

Certifications

The paper trail behind the work

AWS Certified Data Engineer Associate certificate

AWS Certified Data Engineer, Associate

Amazon Web Services · 2026

verify on Credly
DevTown backend web development bootcamp certificate

Backend Web Dev, Express and Node

DevTown, GDSC KIIT, AWS CB · 2023

view certificate

Client work

Software a real business runs on

A small manufacturing business in India runs its billing and books on tools I built. A quoting and invoice app with PDF and Excel export, and a ledger with a calendar view that syncs across devices. Both are in daily use. The code stays private, and a written case study with screenshots is available on request.

ask for the case study

Billing and quoting tool

Invoice maths, print history, PDF and Excel export

Expense ledger

Calendar view, category colours, cloud sync across devices

Hosting and auth

Deployed on Cloudflare with password protected access

Writing

Notes from the build log

Three models are better than one

Why my intrusion detection capstone needed an Isolation Forest, a Random Forest and an autoencoder to reach 0.95 F1 on real attack traffic.

read the post

Contact

Got a hard problem in data?

Email me

vikrantsharma892@gmail.com