Hi, I'm Lookinder Kumar
Building intelligent systems at the intersection of AI, data, and regulated industries.
About Me
AI Engineer building intelligent systems at the intersection of AI, data, and regulated industries.
Lookinder Kumar
AI Engineer & MSc Student
My Story
My work sits at the intersection of machine learning, explainable AI, and real-world financial systems. I build things that are rigorous enough to publish and practical enough to deploy.
Peer-Reviewed Publications
Portfolio Projects
Years ML/AI Experience
MSc In Progress
Interests
What drives me beyond the code
“The goal is to turn data into information, and information into insight.”
— Carly Fiorina
Skills & Tech Stack
The tools and technologies I work with daily
Programming
AI & Machine Learning
LLM Engineering
Data Engineering
Cloud & Big Data
Visualisation & Reporting
Projects
A showcase of my data science, AI, and ML work
Adversarially Robust XAI for Fraud Detection
Full dissertation investigating how XGBoost fraud detection models fail under adversarial attacks (FGSM, PGD, HopSkipJump) and how SHAP explanations invert under those attacks. Mapped to EU AI Act 2024 compliance.
Real-Time Fraud Detection — SWIFT/SEPA Payments
End-to-end pipeline detecting fraud in high-value cross-border SWIFT and SEPA transactions. Hybrid ML detection with SHAP-based reason codes, FastAPI serving, Kafka streaming, and live Streamlit dashboard.
AI-Powered FinTech Market Intelligence
IEEE-format research paper applying K-Means clustering, ARIMA forecasting, and country-level benchmarking to the European FinTech ecosystem. Forecasts funding stabilising at ~$241M/year.
Real-Time Big Data Streaming Pipeline
Kappa-style big data architecture for Transport Infrastructure Ireland M50 traffic data. Emulates real-time streams from CSV, ingests to Apache Kafka, processes with PySpark Structured Streaming, persists to Cassandra.
Diamond Price Prediction & Cut Classification
End-to-end data science pipeline in R on 50,000+ diamond records. Multiple linear regression (Adjusted R² = 0.9207). Cut quality classification: kNN (66%), C5.0 Decision Tree (76.14%), ANN (74.37%).
Brain Tumor Detection & Segmentation — SIYO
IEEE ASPCC 2024 peer-reviewed publication. Proposes the SIYO scheme integrating Meta SAM with YOLOv9 for MRI brain tumour detection. mAP@0.5 = 0.947, accuracy = 0.94.
InfraOS — AI-Native Construction Management
An AI-native SaaS platform for construction project management. Uses LangChain, LangGraph, and the Claude API to automate scheduling, risk flagging, and stakeholder reporting.
Resume
My professional journey and qualifications
Work Experience
Data Science Intern
Aug 2024 — Nov 2024Infinite Computer Solutions · Noida, India
- ›Built Power BI KPI dashboards improving reporting turnaround by 25%
- ›Automated variance analysis (baseline vs actual) using Python/SQL; reduced manual reporting by 8 hrs/week
- ›Supported change control, maintained RAID logs, produced client-ready governance reports
Education
MSc Big Data Management & Analytics
Jan 2025 — Jun 2026Griffith College Dublin
Projected First Class Honours. Key modules: Big Data, Cloud Platforms, Information Retrieval, Parallel & Distributed Programming, Data Mining, Statistics for Data Science, Research Methods, Data Visualisation & BI. MSc thesis: Adversarial Robustness & SHAP Stability in Fraud Detection under EU AI Act 2024.
BTech Computer Science & Engineering
Graduated May 2024C.V. Raman Global University, India
CGPA 8.50/10 — First Class Distinction. Focus on algorithms, data structures, machine learning, and software engineering fundamentals.
Volunteering
Secretary
Jan 2025 — PresentErasmus Student Network (ESN) — Griffith College Dublin
Representing 100+ international students. Improved cross-cultural engagement by 40%. Managing communication, event logistics, and collaboration with ESN Ireland national board.
Blog
Thoughts on data science, ML engineering, and AI research
Adversarial Attacks on Fraud Detection: What My Thesis Found
My MSc thesis set out to answer a dangerous question: what happens to XGBoost fraud detection models and their SHAP explanations when a sophisticated adversary deliberately crafts transactions to evade detection? The answer was worse than expected.
Why SHAP Explanations Break Under Adversarial Pressure
The EU AI Act classifies fraud detection systems as high-risk AI. Article 13 requires meaningful explanations. But what if the explanations themselves can be manipulated by the very adversary you're trying to detect? This is the regulatory gap my research addresses.
SWIFT & SEPA Payments: How AI Can Catch What Rules Miss
Rule-based systems flag what they've seen before. Machine learning models catch what rules miss. But neither alone is enough for high-value cross-border payments where milliseconds and millions are both at stake. Here's how I built a hybrid detection pipeline.
Get in Touch
Whether you're a recruiter, a PhD supervisor, or building something in AI — I'd love to connect.