Aspiring Machine Learning Engineer. B.E. AI & ML at Mysore University · GPA 9.31/10 · recently shipped a production-ready Customer Churn Intelligence System with LightGBM and Streamlit.

"Building cool stuff — one algorithm at a time."
Abhishek Rai A, a B.E. AI & ML student at Mysore University School of Engineering, entering my fifth semester with a 9.31/10 GPA. Originally from Mangalore, now in Mysuru.
I find patterns in mathematics and elegance in algorithms. I speak Tulu (mother tongue), Kannada, Hindi and English. I recently completed a rigorous 4-week, roughly 120-hour Core AI & ML internship at Bluemind Solutions Pvt. Ltd. (Aug–Sep 2026), where my capstone was a leak-free Customer Churn Intelligence System on IBM Telco data.
I compared Logistic Regression, Random Forest, XGBoost and LightGBM, tuned the decision threshold for business cost, added SHAP-style analysis, and shipped a live Streamlit dashboard. LightGBM reached 0.849 CV ROC-AUC and 82% recall on churners, with an estimated ~$179k in annual recoverable revenue. One Piece bias included.
Bluemind Solutions capstone: leak-free tabular ML, business threshold tuning, and a live Streamlit dashboard on IBM Telco data.
Ongoing certification via DRISHTI CPS (IIT Indore) — ML fundamentals, workflows, applied AI.
Learning the next layer of ML engineering deliberately — strengthening model serving and deep-learning fundamentals without overclaiming production experience.
Core weapons in the arsenal — AI/ML + math foundations. I vibe-code web (HTML/CSS/JS/React) with AI as copilot. Focus stays on the algorithms.
Not hardcoded — pulled live from the GitHub API every 10 minutes. What you see is what I actually shipped.
Click any card for screenshots and the live site — or read the full case study for the problem, constraints, trade-offs and what actually shipped.
Recruiter in a hurry? Skip the scroll — ask the AI trained on my resume and projects. Answers stay grounded in my actual background.
Hi — I'm the assistant trained on Abhishek's resume and projects. Ask me anything about his background, projects, or availability.
A separate world I built
The Grand Line was too big for one portfolio section—so I built it its own world: a scroll-driven 3D shelf of 60 titles, with mood-based AI picks, random discoveries, and live visitor recommendations.