// PORTFOLIO — HUBLI, KARNATAKA, INDIA

Aishwarya Hulmani

 

I move between two disciplines by choice: building full-stack applications end to end on the MERN stack, and building computer vision and NLP models that need to actually work on messy real-world data.

Say hello ↗ LinkedIn View work
4College projects
3Industry simulations
92%ML model accuracy

01 · About

Builder, both sides of the stack

I graduated with a BCA from KLE Technological University in 2025, and spent most of that time building things rather than just studying them — expense trackers with real authentication logic, a job-aggregation tool that scrapes and parses actual postings, and object-detection models trained on underwater imagery that doesn't behave like a clean textbook dataset.

I'm comfortable owning a project from database schema to trained model to working interface, and I'm currently looking for a software engineering or ML/data role where I can keep doing exactly that.

Quick facts

  • BCA, KLE Technological University — 2025
  • GPA 7.26 / 10
  • Based in Hubli, Karnataka, India
  • English · Hindi · Kannada

02 · Skills

What I build with

💻

Web Development

MongoDBExpress.jsReact.jsNode.jsREST APIsRBAC
🤖

Machine Learning

scikit-learnYOLOVGG16NLPTF-IDF
📊

Data Analysis

PandasNumPyMatplotlibEDA
🗄️

Databases

MongoDBMySQL
☁️

Cloud & Tools

AWSOCIGitVS Code
🧠

Languages

PythonJavaScriptSQL

03 · Projects

Things I've built

💸

Finex — Full-Stack Expense Tracker

MongoDBExpressReactNode
  • Full-stack expense app end to end across a modular MERN architecture, from schema to dashboard.
  • Secure authentication and role-based access control via custom REST APIs.
  • Real-time dashboards, deployed on MongoDB Atlas.
🧭

Smart Job Aggregator

MERNWeb ScrapingPDF.js
  • Aggregates listings from LinkedIn, Naukri, and Indeed via automated Cheerio scrapers.
  • Resume-parsing module with PDF.js plus a candidate-to-job matching algorithm.
  • Built around MERN best practices across every layer.
🌊

Underwater Garbage Detection

YOLOVGG16OpenCV
  • Marine-waste detection pipeline combining YOLO and VGG16 for multi-class object detection.
  • Preprocessing and augmentation to handle inconsistent lighting and water clarity.
  • Evaluated with mAP, precision, and recall.
📰

News Article Categorization

scikit-learnNLPTF-IDF
  • Multi-class text classification with Naive Bayes and Decision Trees.
  • TF-IDF vectorization plus an NLP preprocessing pipeline.
  • 92% accuracy, 0.89 F1-score, confusion-matrix validated.

04 · Experience

Virtual experience programs

JUNE 2025

Operations Job Simulation

Goldman Sachs, via Forage

Simulated fund transfer processing and risk reporting analysis for a global markets operations team.

JUNE 2025

Data Forensics Program

Deloitte, via Forage

Worked through data forensics tasks, including forensic data analysis and fraud-pattern investigation techniques.

SEPTEMBER 2025

Software Engineering Virtual Experience

JPMorgan Chase & Co., via Forage

Completed software engineering tasks covering debugging, feature coding, and technical documentation modeled on JPMorgan Chase's engineering workflow.

05 · Education

Education & certifications

Bachelor of Computer Applications (BCA)

KLE Technological University, Hubli — 2022–2025 — GPA 7.26 / 10

Oracle Cloud Infrastructure (OCI) Data Science Professional — 2025
AWS Fundamentals of Machine Learning
AI Engineer — One Roadmap
Big Data Tools — Coursera
Career Essentials in Data Analysis — Microsoft & LinkedIn
ABCs of Malware — Great Learning