← Back to portfolio

Abhishek Rai A

Aspiring Machine Learning Engineer · Mysuru, Karnataka, India

Profile

B.E. student in Artificial Intelligence & Machine Learning entering the fifth semester with a 9.31/10 GPA. Recently completed a rigorous 4-week (~120 hour) Core AI & ML internship at Bluemind Solutions, where he shipped a Customer Churn Intelligence System with LightGBM and Streamlit.

Education

B.E. — Artificial Intelligence & Machine Learning2024 — 2028
Mysore University School of Engineering · GPA 9.31 / 10 · Entering V Semester
Certification — AI & Data ScienceOngoing
DRISHTI CPS Foundation, IIT Indore · ML fundamentals, workflows, applied AI

Skills

Proficient
PythonNumPyPandasScikit-learn (Pipelines, ColumnTransformer, CV)GitLinear AlgebraCalculus
Intermediate
XGBoostLightGBMMatplotlib / SeabornStreamlitSQLDSAProbability & Statistics
Learning / Basics
PyTorchFastAPIDiscrete Mathematics
Web
ReactTypeScriptTanStack StartSupabaseTailwind CSS

Projects

A production-ready customer churn intelligence system completed as the capstone for Bluemind Solutions' Core AI & ML Internship.

  • 0.849 CV ROC-AUC
  • 82% churner recall
  • ~$179k est. annual recovery
Python · NumPy · Pandas · Scikit-learn · XGBoost · LightGBM · Streamlit

An anonymous-but-accountable grievance channel for a university, where a complaint only becomes official once enough peers back it.

  • 1,159 verified student IDs
  • 0 identities exposed to client
  • SSR edge-rendered
TanStack Start · React 19 · Supabase (Postgres, RLS) · Cloudflare Workers · TypeScript

A sorting visualiser built for people who already know the pseudocode and still can't feel why quicksort degrades.

  • 5 algorithms, step-for-step
  • ~60fps on 200-element arrays
  • 1 embeddable /embed widget
React · TypeScript · Web Audio API · Framer Motion

The smallest thing I could build that makes O(log n) obvious in one look.

  • 0 dependencies / frameworks
  • log₂n steps shown live
  • <50KB shipped to the browser
JavaScript (ES modules) · CSS · Netlify

Three console programs written to make OOP stop being vocabulary and start being a tool.

  • 3 programs, one shared core
  • 100% input paths validated
  • C++17 standard, no libraries
C++17 · STL · g++

Languages

Tulu (mother tongue) · Kannada · Hindi · English

How I work

Night owl, ships fast, iterates in public. I use AI copilots deliberately for the web layer so my own hours go into the algorithms and the maths. Goal: one small ML system shipped every month.