Parth Jadhav

Software Engineer - Building practical AI products and clean web experiences.

I enjoy turning research-heavy ideas into usable products with strong user experience and reliable system behavior.

From my experience, I have gained strong skills in API design, backend reliability, full-stack delivery, and shipping AI/ML features that are practical, testable, and production-focused.

Parth Jadhav

Work Experience

Unique School App

Jan 2026 – June 2026

Hybrid

  • Migrating a school-management platform from PHP to a modern Node.js stack.
  • Designed REST APIs for core modules, including testing and end-to-end integration.
  • Turned frontend prototypes into production UI and kept backend services reliable.
  • Node.js
  • Next.js
  • REST API
  • React
  • JavaScript
  • MySQL

Tech Mahindra

June 2024 – Aug 2024

On-site

  • Fine-tuned a Whisper speech-to-text pipeline for Local Bahasa translation across accents.
  • Built semantic search for an internship portal with sentence-transformers and ChromaDB.
  • Contributed to Malay and Sanskrit LLM work and integrated models into web interfaces.
  • Python
  • Whisper
  • PyTorch
  • ChromaDB

Skills

  • Python
  • JavaScript
  • React
  • Next.js
  • Node.js
  • Flask
  • FastAPI
  • TensorFlow
  • PyTorch
  • Scikit-learn
  • Pandas
  • NumPy
  • REST API
  • Git
  • Whisper
  • NLP

Projects

Adaptive Medical Tokenizer

A compact tokenizer tuned for real-world medical text noise.

Built a vocabulary-adaptive tokenizer for clinical NLP that learns new medical terms during inference, cutting token count 50–70% versus GPT-2 while keeping cardiology meaning intact.

  • Python
  • NLP
  • Streamlit

Relay

A native Windows 11 command palette and productivity launcher.

WinUI 3 tray app with fuzzy app search, clipboard history, snippets, window tiling, file search, and AI actions. Opens with a global hotkey.

  • C#
  • WinUI 3
  • .NET

Parkinson's Disease Detection

ML workflow for screening Parkinson’s disease from vocal features.

Trained XGBoost and Random Forest classifiers on speech-based audio metrics. Published in IEEE Punecon’24 and received the best research paper award.

  • Python
  • XGBoost
  • Scikit-learn

Typely

Ubuntu tray app for fast speech-to-text transcription.

Privacy-first voice typing with local Whisper, global hotkeys, tray controls, clipboard/cursor paste, and a Debian installer. No cloud APIs.

  • Python
  • Linux
  • Whisper

Get in Touch

Contact me at jadhavparth2004@gmail.com.