builder

Software engineer · Pune, India

Harsh Mehta

SDE Intern at Quantifai, Google Summer of Code 2026 contributor with Liquid Galaxy, and final-year B.Tech IT student at VIT Pune.

Current
Quantifai · SDE Intern
Open source
Google Summer of Code 2026 · Liquid Galaxy
Education
VIT Pune · 8.95 CGPA
Scroll to explore

About

Who am I?

Curiosity is usually where the useful work starts.

I'm driven by curiosity and a deep appreciation for the beauty in complexity. I love understanding how things work, which has led me to explore diverse fields, from full-stack development to cybersecurity and AI agents.

I'm a GSoC 2026 contributor working on the Liquid Galaxy project, where I'm building Local AI with Gemma: an AI agent server enabling natural-language control of a multi-screen globe setup.

I love to study system design, collaborate with developers worldwide, and apply what I learn to real problems.

VIT Pune · B.Tech in Information Technology · CGPA 8.95

500+
DSA problems solved
3
Research papers
4x
National hackathon winner
60+
Students mentored

Skills

Tools I work with

Languages

TypeScriptJavaScriptPythonSQLC++Java

Backend

RabbitMQPrefectFastAPINode.jsExpress.jsDesign PatternsConcurrency

Frontend

ReactNext.jsFlutter

Data & tools

PostgreSQLMongoDBMySQLClickHouseGitKali Linuxn8n

DevOps & cloud

DockerCI/CDGitHub ActionsAWSAzureGCP

AI / ML

Machine LearningLLMOpsRAGHermes AgentPrompt EngineeringAI Security

Experience

Where I've been

Jul 2026 — Present

SDE Intern

QuantifaiEnterprise financial technology

Prefect · FastAPI · Next.js · PostgreSQL · Docker · RabbitMQ

  • Owning backend systems processing 11Cr+ transactions monthly across production pipelines serving enterprise clients.
  • Redesigned core processing logic to cut end-to-end latency from 8–9 seconds to 900ms by batching I/O and optimizing API endpoints.

May 2026 — Aug 2026

Contributor · Liquid Galaxy

Google Summer of CodeGoogle-incubated open-source project

Local & cloud LLMs · Agentic pipelines · Hermes Agent · Shell scripting

  • Selected among 23,000+ global applicants for GSoC 2026.
  • Built a self-learning Hermes agent profile that converts natural-language inputs into geospatial visualisations and reduced creation time from hours or days to under two minutes.

Sep 2025 — Nov 2025

Intern · Green IT & Sustainability

Tech MahindraSustainable infrastructure research

ML forecasting · DNS infrastructure

  • Built an ML-based DNS optimization system for energy efficiency.
  • Reduced redundant lookups by 16% and estimated infrastructure energy draw by 10–12%.

Jun 2025 — Jul 2025

Software Intern

ITARIUM Technologies IndiaApplication security

Burp Suite · Penetration testing · OWASP Top 10

  • Performed black-box and grey-box VAPT on production web apps and APIs; reported SQL injection, broken authentication, and authorization flaws with ranked remediation reports.

Projects

Things I've built

Selected work across full-stack systems, security, AI, and open source.

01

α SHUP — Retail Operations & CRM

A production retail CRM for Kalpana Industries in Amravati, replacing paper workflows across inventory, billing, customer management, referrals, and rewards.

MERNRBACREST APIsProduction
Live across all stores
02

Hermes Agent — Google Summer of Code 2026

A self-learning agent profile for Liquid Galaxy that converts natural-language requests into geospatial visualisations, KML layers, and guided tours with text-to-speech.

Hermes AgentAgentic AILiquid GalaxyOpen Source
View GSoC project
03

Ship Rust Detection & Analysis

An AI corrosion-detection backend for Creatum GmbH, Hamburg, classifying rust from ship imagery with a self-training pipeline built for the absence of a public ship-part dataset.

PythonDeep LearningComputer Vision
Creatum GmbH
04

Open Source Contributor

Contributions across OWASP, Google Liquid Galaxy, and other open-source projects, including a Hacktoberfest SuperContributor finish.

Open SourceOWASPLiquid Galaxy
Ongoing
05

RAG based PDF Search Tool

An AI-powered document search tool using retrieval-augmented generation for focused querying across PDF content.

RAGPythonAI / ML
View repository
06

Realtime DDoS Detection System

Live network-traffic analysis that identifies DDoS behavior and surfaces alerts as attacks develop.

CybersecurityNetworkingPython
View project
07

Network Intrusion Detection System

A machine-learning NIDS that classifies network packets and detects malicious intrusion patterns.

Machine LearningSecurityNIDS
View project
08

Skin Cancer Classification

A deep-learning web application that classifies skin-lesion images to support early detection research.

Deep LearningCNNHealthcare
Research project

Publications & patent

Research in print

  1. P01

    Smart Curve Guardian: Real-Time Vehicle Threat Detection

    6th International Conference on Recent Trends in Communication & Intelligent Systems (ICRTCIS)

    Also granted as a patent in the Republic of South Africa (CIPC), Application No. 2025/10085, July 2026; co-inventor with the VIT Pune team.

  2. P02

    DoS Attack Detection Using ML

    International Conference on Electronics, Computing & Communication Automation Technologies (ICEC2NT)

  3. P03

    The Future of Skin Cancer Detection: ML and CNNs

    5th International Conference on Data Intelligence & Cognitive Informatics (ICDICI)

Achievements

Milestones & recognition

AIR 47

AICTE Pentathon 2025

Government of India National Cybersecurity Challenge · 22,000+ participants

4x

National Hackathon Winner

Including CRESA Canvas and i-Smart India Hackathon · Top 15 of 900+ teams

500+

DSA Problems Solved

Across LeetCode and other competitive-programming platforms

60+

Students Mentored

Project Head, IEEE Student Branch VIT Pune · 2024–2026

4700+

LinkedIn Community

Learning, building, and sharing in public

Blogs

Notes from the rabbit hole

Browse all blogs ↗

Contact

Let's talk.

Send an email