Available for opportunities

Kevalya Shah

Software Engineer · AI & ML · based in India. I build scalable web systems and apply machine learning to solve real-world problems.

3+
Internships & roles
10+
Shipped projects
6
Languages used

About

I'm a software engineer who enjoys living at the intersection of full-stack engineering and machine learning. I've fine-tuned speech and translation models, built data pipelines that move fast, and shipped end-to-end web platforms. I care about clean systems, measurable impact, and learning something new on every project.

Experience

Application Developer @ Larsen & Toubro

Jan 2026 — Present

Full-stack development driving feature improvements across the platform.

Full-Stack Web Enterprise

AI Intern @ Intelehealth

May 2025 — Jul 2025

Researched and fine-tuned speech-to-text and translation models, improving transcription accuracy and handling large-scale audio datasets.

NLP Speech PyTorch

Software Developer Intern @ Kellogg Brown & Root

May 2024 — Jul 2024

Built Python data pipelines that cut processing time and improved ML model accuracy on new datasets.

Python Data Pipelines ML

Skills

Languages

Python C C++ Java TypeScript JavaScript C#

Web & Frameworks

React Next.js Node.js MERN Stack Tailwind .NET

Data & ML

TensorFlow PyTorch Machine Learning Deep Learning Pandas

Databases

MongoDB SQL PL/SQL

Foundations

Data Structures & Algorithms System Software OOP

Projects

Explainable deep-learning model that classifies vehicle damage severity, with Grad-CAM heatmaps so predictions can be visually justified.

Python Deep Learning · Grad-CAM

Final-year project that models race conditions to recommend optimal pit-stop strategy using data-driven decision making.

Python AI · Optimization

End-to-end workflow platform handling the full employee-transfer lifecycle, from request to approval.

HTML Full-Stack · Workflow

A matching engine for trading orders, built to explore low-latency order-book design and matching logic in Java.

Java Java · Systems

A voice assistant integration demonstrating real-time speech interaction with large language models.

JavaScript Voice · LLM

A deep-learning model trained to recognise my own handwriting — a hands-on dive into CNNs and dataset building.

Jupyter Notebook CNN · Deep Learning

Project

HTML 1

No description provided.

HTML

No description provided.

JavaScript

A simple gym trainer application which can be used to learn different workout techniques or schedule your workout as per needs.

TypeScript

A simple Todo app created using C# and .NET 10 with Onion architecture implementation

C#

No description provided.

Java

No description provided.

Python
View all on GitHub →

Achievements

Showcased inter-school winning robotics project at EXPO 2020 Dubai
Runner-up at Mindbend, NIT Surat
Active competitive programmer on LeetCode

Get in touch

Have a project, role, or idea in mind? Drop a note — I'll get back to you.