model.predict(gesture)
// 92.0% accuracy

SELECT * FROM invoices
WHERE semantic_match(query)

class BiLSTM(nn.Module):
def forward(x):

$ pytest --cov
✓ 2,500+ tests passed

OPEN TO OPPORTUNITIES — 2026

Hi, I'm Kashish Garg

Software Engineer • AI Researcher • Machine Learning Enthusiast

I build intelligent software that combines scalable engineering with cutting-edge AI. From enterprise backend systems to published AI research, I enjoy solving meaningful problems that create real-world impact.

Download Resume

SCROLL TO EXPLORE

$whoami

kashish_garg

$cat role.txt

Final-year ECE student building AI systems

$ls interests/

computer-vision/ nlp/ edge-ai/ accessibility-tech/

$status

open to full-time roles — 2026

$ ▍

About Me

Engineering intelligence,
one system at a time.

I'm Kashish Garg, an Electronics and Communication Engineering student about to graduate, passionate about Software Engineering, Artificial Intelligence, Computer Vision, and Natural Language Processing. My work spans production software, AI research, and intelligent applications.

I enjoy transforming ideas into reliable products while continuously exploring emerging technologies through research and hands-on development — from a custom-trained sign-language model to a semantic search system for buried invoices.

0+ LeetCode Problems Solved

0+ Members Led at AIESEC

0+ Unit Tests Shipped

Experience

Where I've been building

A short timeline — the line draws itself in as you scroll.

RECENT

Software Development Engineer Intern

UKG

Reduced infrastructure costs by 20% through automation.
Debugged high-impact production issues and improved observability.
Generated 2,500+ unit tests across Java and Python workflows, improving reliability.

EARLIER

AI Research Intern

Applied Machine Learning Research

Worked on applied machine learning and statistical methods.
Analyzed datasets and evaluated predictive models for real-world applications.
Findings fed directly into later peer-reviewed publication work.

Selected Work

Things I've shipped

Keep scrolling — each project slides in as the last one slides away.

SCROLL TO CONTINUE ↓

ACCESSIBILITY · COMPUTER VISION

Indian Sign Language to Speech Converter

A real-time accessibility solution that reads ISL gestures through a camera and speaks them aloud, built to bridge communication between deaf and hearing individuals.

92% accuracy · 200+ custom-trained gestures · 30-frame sequences

MediaPipeOpenCVTensorFlowPyTorchBiLSTM

NLP · SEMANTIC SEARCH

AI-Powered Email Invoice Search System

An intelligent document retrieval platform that finds invoices and emails using natural language instead of exact keyword matching.

Semantic search over OCR'd emails, no keyword guesswork required

Gmail APITesseract OCRSentenceTransformersSQLiteStreamlit

DESKTOP APP · JUNE 2024

Stichwise & Co

A JavaFX desktop application for digitizing order management for small-scale tailors, with secure role-based logins and an interactive dashboard for tracking measurements and deliveries.

Secure multi-role login · automated email order confirmations

JavaJavaFXMySQL

FULL-STACK · AUGUST 2024

Sports Pulse

A full-stack sports event management platform with RESTful APIs, secure user authentication, and a dynamic event tracking system built to handle real traffic.

1,000+ users served · RESTful Node.js APIs

Node.jsAngularJSHTML5CSS3jQueryAJAXJSONMySQL

01 / 04

Publications

Peer-reviewed work

SPRINGER LNCS · PRESENTED AT ICDCIT 2026

A Novel Explainable Multimodal Edge Framework for Stress Detection

This research proposes an explainable multimodal edge-computing framework for stress detection by integrating machine learning with edge computing principles. The work focuses on improving model interpretability while enabling efficient deployment on edge devices for healthcare applications.

0Papers Accepted

0Submissions

2026ICDCIT

Skills

What I bring to the table

A categorized snapshot of my technical toolkit.

LNGLanguages

CC++PythonSQLJava

DBDatabases & Cloud

MySQLGrafanaGCPPostgreSQL

WEBWeb Technologies

HTMLCSSJavaScriptBootstrapAJAXNode.js

MLData Science & ML

NumPyPandasPyTorchTensorFlowKerasDeep LearningComputer VisionNLPReinforcement Learning

Leadership

Beyond the codebase

AIESEC · VICE PRESIDENT

Leading people, not just projects

Led 100+ members, improved onboarding efficiency by 35%, coordinated 10+ international cultural exchange programs, and managed cross-functional teams while delivering impactful volunteer experiences.

0+ Members Led

0% Onboarding Efficiency

0+ Exchange Programs

Contact

Let's build technology
that matters.

Open to full-time roles, research collaborations, and interesting freelance work.

Say hello

Whether it's a role, a project, or just a question about a paper — my inbox is open.

bansalk5000@gmail.com