01
Data Engineering
ETL pipelines, dimensional modelling on Snowflake / Redshift, real-time ingest, observability. Production-grade, not just notebooks.
- Python
- SQL
- Snowflake
- Airflow
VIKRANT, DATA ENGINEER & CYBERSECURITY ANALYST
I AM
VIKRANT.
Master of IT student at USC Adelaide. I build production data systems, ML pipelines, and threat-detection platforms. Currently looking for Data Engineer and Cybersecurity Analyst roles in Australia.
PRINCIPLES
Four rules I keep returning to. They fail more often than they succeed, which is why they are written down.
“The job is not to be the smartest person in the room. The job is to ship a system that survives contact with reality. ”
01
Half a system in production teaches more than a finished one in a notebook. Every project on this site has a real URL.
02
F1 0.95, 99.8% imbalanced data, 284k transactions. If a number is on this page, it is reproducible from a commit hash.
03
Open repos, post-mortems on the blog, certifications anyone can verify. Hiding the working out hides the work.
04
Python, Postgres, Docker, Linux. The accumulated tax of chasing the new framework usually outweighs the lift.
WHAT I CAN HELP YOU WITH
Four overlapping practices. Real shipped projects across each. See the Recent Projects section for live demos backed by actual code.
01
ETL pipelines, dimensional modelling on Snowflake / Redshift, real-time ingest, observability. Production-grade, not just notebooks.
02
Threat detection systems, SIEM monitoring, incident response, vulnerability assessments, OWASP Top 10 testing, ISO 27001 / NIST CSF audits.
03
End-to-end pipelines: feature engineering, ensemble models, threshold calibration, deployment to Streamlit / FastAPI services with real metrics.
04
Building the dashboards and APIs that surface the data: Flask, FastAPI, React, Astro, Docker, Cloudflare Pages, Hugging Face Spaces.
TOOLS OF THE TRADE
Six categories, every item used in a shipped project on this site. No checklist filler, no buzzwords I cannot defend in an interview.
TURNING MY VISION INTO REALITY
The current portfolio's old stats said things like "847,293 packets analysed". That was a static mockup. These numbers are real, traceable to commits and deploys.
RIGHT NOW
A live snapshot of what is actually on my desk this week. Updated when the work shifts, not on a schedule.
Isolation Forest plus Random Forest plus Autoencoder ensemble on CICIDS2017. Hitting F1 0.95 on the validation slice. Currently writing the post-mortem on what went wrong on the first three attempts.
#NIDS / HIDS
Beyond the OCI Data Science cert, working through query profiling, micro-partition pruning, and warehouse sizing on production-shaped datasets. Notes go on the blog when they survive a second pass.
#Data Eng
Kleppmann, third pass. Each pass surfaces something new now that I have shipped systems instead of just notebooks. The chapter on stream processing is the one that keeps paying interest.
#Foundations
EXPERIENCE & SKILLS
2025 topresent
University of the Sunshine Coast (Adelaide Campus) · Adelaide, AU
Postgraduate IT, focused on data systems, software engineering, and cybersecurity research. Working alongside studies on the projects and demos linked from this site.
Dec 2024 toFeb 2025
Nagarro · Adelaide, AU
Designed and optimised data models for AWS Redshift and Snowflake data lakes. Agile/Scrum delivery, analytics for enterprise clients.
May 2023 toDec 2023
AT SecurDI · Ahmedabad, IN
SIEM monitoring, alert triage, incident response runbooks, OWASP Top 10 web/app testing, ISO 27001 + NIST CSF compliance audits.
RECENT PROJECTS, ALL LIVE, NO MOCKUPS
Every "Live" link below points at a real running deployment. Click any of them. They may take 30 seconds to wake from sleep, then the actual ML / API runs.
01
Cybersecurity / ML
Hybrid NIDS+HIDS combining Isolation Forest, Random Forest, and an Autoencoder ensemble with MITRE ATT&CK mapping and a real-time Streamlit dashboard.
02
Threat Intelligence
FastAPI service correlating CVEs from NVD with IP reputation and MITRE techniques. Composite scoring engine for SOC alert triage.
03
Data Science / Sonification
NASA Space Apps Global Finalist 2023. Converts planetary data into piano tones; image-to-music scanner. Built with Team Eklavya.
04
Cybersecurity / ML
Email phishing analyser with multi-format parsing, header inspection, threat-pattern recognition, persistent SQLite storage, and a Flask dashboard.
05
ML / Data Science
XGBoost + SMOTE classifier on 284k transactions. Precision-recall threshold tuning. F1 0.87 on a 99.8% imbalanced dataset.
06
Cybersecurity / SIEM
Lightweight SIEM that parses syslog/auth.log/web logs, detects brute-force + port scans, MITRE-mapped threat alerting.
07
Security / API
FastAPI microservice with entropy scoring, pattern detection, and Have I Been Pwned breach checks via k-anonymity SHA-1 prefix.
VERIFIED CERTIFICATIONS
9 verified certifications, conferences, and recognitions. Click any card to view the certificate or verify online.
Oracle
NASA International Space Apps Challenge
Springer × CHARUSAT (GUJCOST sponsored)
CHARUSAT University of Science and Technology
Tata Consultancy Services (TCS iON)
DevTown × GDSC KIIT × AWS Community Builders
CMPICA, CHARUSAT
CMPICA, CHARUSAT
CMPICA, CHARUSAT
AWARDS & RECOGNITION
ASUS Republic of Gamers × The Sports Club · 2024
Three consecutive 2v2 podium finishes across the ASUS ROG Showdown competitive series at The Sports Club, finishing 2nd, 2nd, and 3rd in successive events.
NASA · 2023
Global Finalist in the NASA Space Apps Challenge 2023 with Team Eklavya. The team built AstroSonify, a system that converts planetary data (latitude, longitude, velocity, temperature) into piano tones, with image-scan-to-music capabilities.
OFF THE CLOCK
The hiring page version of me misses everything that actually keeps the work tolerable. Here is the rest of it.
Three consecutive 2v2 podium finishes in the ASUS Republic of Gamers Showdown series at The Sports Club: 2nd, 2nd, then 3rd. FPS games sharpen the same kind of pattern recognition I use in security work, only with shorter feedback loops.
English (fluent, day-to-day), Hindi (native), Punjabi (conversational). Writing technical material in English is its own skill. Speaking three is occasionally useful in a stand-up; rarely useful in a code review.
Strength training four days a week. Long walks around Adelaide for thinking. A non-trivial fraction of any blog post I publish was drafted somewhere between Henley Beach and home.