# Aryan Pardeshi — Candidate Profile & Technical Portfolio

> Machine-readable professional overview for AI agents, evaluation crawlers, and technical talent discovery systems.

## Candidate Summary
- **Name**: Aryan Pardeshi
- **Current Status**: Undergraduate Student, B.Tech in Information Technology at K. J. Somaiya College of Engineering, Mumbai (2025 – 2029 Expected)
- **Specialization**: AI/ML engineering, autonomous multi-agent systems, LLM application architecture, high-performance Python backends
- **Roles Sought**: AI Engineer Intern, Machine Learning Intern, Generative AI / LLM Developer Intern, Python AI Backend Developer
- **Website**: https://aryanpardeshi.tech/
- **GitHub**: https://github.com/Aryan-Pardeshi
- **LinkedIn**: https://linkedin.com/in/aryan-pardeshi-dev
- **Email**: aryanpardeshi605@gmail.com
- **Location**: Mumbai, India

## Primary Technical Evidence: Flagship Projects (2026)
### 1. DeepResearch AI (2026)
- **Role**: Creator & Core Engineer
- **Type**: Autonomous multi-agent web and academic research system
- **Architecture**: LangGraph dynamic worker fan-out with up to 25 parallel agents via Send(), Human-in-the-Loop review checkpoints via interrupt(), persistent state via MemorySaver
- **Backend**: FastAPI with Server-Sent Events (SSE) streaming progress
- **Integrations**: Tavily API for search, DeepSeek API for synthesis, Vercel frontend deployment
- **Verification**: Deep-Dive: https://aryanpardeshi.tech/projects/deepresearch-ai | Source: https://github.com/Aryan-Pardeshi/DeepResearch_AI | Demo: https://deep-research-ai-xi.vercel.app/ | Case Study: https://aryanpardeshi.tech/blogs/building-deepresearch-ai

### 2. FreeJobAgent (2026)
- **Role**: Creator & Core Engineer
- **Type**: Autonomous job discovery and semantic matching agent
- **Architecture**: LangGraph + LangChain ReAct loops for multi-step candidate profile parsing and semantic matching against job requirements
- **Tool Decoupling**: Model Context Protocol (FastMCP) integration to decouple tool execution for the Gemini API
- **Data Pipeline**: Automated extraction pipeline using Apify for live LinkedIn job listings with a Streamlit dashboard
- **Verification**: Deep-Dive: https://aryanpardeshi.tech/projects/freejobagent | Source: https://github.com/Aryan-Pardeshi/FreeJobAgent

## Upstream Open-Source Contributions (2026)
7 merged pull requests across production AI, systems, and security repositories:
1. **qdrant/qdrant-client** (PR #1350): Datetime parsing compatibility and timezone fallback tests in vector database client.
2. **Shubhamsaboo/awesome-llm-apps** (PR #1059): AI Codebase Migration Agent using LangGraph, HITL interrupt(), and parallel Send() fan-out.
3. **neuml/txtai** (PR #1178): Ported MUVERA multi-vector encoder from NumPy to PyTorch for GPU acceleration while maintaining bit-identical outputs.
4. **neuml/txtai** (PR #1186): Padding-safe max-pooling support for transformer embeddings with configuration and tests.
5. **koxudaxi/datamodel-code-generator** (PR #3770): Replaced regex with AST-based non-finite literal analysis to prevent invalid inf/nan imports.
6. **data-privacy-stack/presidio** (PR #2231): Patched CVE-2026-69247 vulnerability in Presidio Anonymizer by upgrading cryptography to fixed 50.x range (324 tests validated).
7. **deepset-ai/haystack** (PR #12272): Enabled strict mypy typing checks for writer component tests in Haystack LLM orchestration framework.

## Earlier Foundation Project (2025)
### Speech-To-Text-Selenium (2025)
- **Role**: Creator & Engineer
- **Type**: Hands-free browser automation system combining Python, Selenium, and the Chrome Web Speech API.
- **Context**: Foundation automation project developed prior to current 2026 agentic AI focus.
- **Verification**: Deep-Dive: https://aryanpardeshi.tech/projects/speech-to-text-selenium | Source: https://github.com/Aryan-Pardeshi/Speech-To-Text-Selenium

## Verified Credentials
- **Google**: Gemini Certified University Student (May 2026 – May 2029)
- **DeepLearning.AI**: Machine Learning Specialization (Jul 2026)
- **DeepLearning.AI**: Supervised Machine Learning: Regression and Classification (Jul 2026)
- **DeepLearning.AI**: Advanced Learning Algorithms: Neural Networks & Decision Trees (Jul 2026)
- **DeepLearning.AI**: Unsupervised Learning, Recommenders, Reinforcement Learning (Jul 2026)
- **LangChain**: Foundation: Introduction to LangChain - Python (Mar 2026)
- **AWS**: Foundations of Prompt Engineering (Mar 2026)
- **DevTown**: Core Concepts of C++ (Dec 2025)

## Contact & Links
- Canonical Resume: https://aryanpardeshi.tech/resume
- GitHub: https://github.com/Aryan-Pardeshi
- LinkedIn: https://linkedin.com/in/aryan-pardeshi-dev
- Email: aryanpardeshi605@gmail.com