I’m an AI-focused Software Engineer specializing in LLM applications, agentic systems, and backend engineering. I build production-grade systems that combine machine learning, multi-agent architecture, and scalable backend services to solve real-world problems — from intelligent travel systems to AI-driven content pipelines.
I thrive at the intersection of engineering + applied AI and enjoy turning research-grade ideas into deployable systems.
🌐 Portfolio: (https://arya-pathrikar-portfolio.vercel.app) 🔗 LinkedIn: (https://www.linkedin.com/in/arya-pathrikar/) 📩 Email: arya.pathrikar@gmail.com
Java, Python, Spring Boot, FastAPI, MongoDB, MySQL, PostgreSQL, Bash
React, Tailwind CSS, HTML, CSS, TypeScript, Node.js
Git, Docker, Kubernetes, AWS, Linux, Agile/Scrum, REST APIs
Pandas, PyTorch, NumPy, OpenCV, Hugging Face Transformers, LLMs, BERT, LangChain
Tech: Gemini LLMs, Google ADK, MCP, Multi-Agent Systems
Built a multi-agent system using Google ADK + Gemini LLM to extract landmarks from Instagram Reels (Vision Agent), resolve locations, and generate personalized travel itineraries, achieving ~92% landmark detection accuracy.
Implemented custom MCP tools, parallel + sequential agents, and session/state management with context-aware memory, improving multi-step reasoning consistency by 40%.
Added production-grade observability (logs, metrics, traces), increasing itinerary relevance and reliability by 30% through performance tuning.
Tech: Python, XGBoost, Random Forest, ML Pipelines
Processed biometric signals (HRV, temperature) from wearable devices with 500+ data points for stress classification.
Compared ML models (Random Forest, XGBoost, SVM) with advanced outlier detection (IQR, Isolation Forest), achieving 92% classification accuracy.
Applied feature engineering and predictive modeling to enable biometric-based stress insights.
Tech: Java, Spring Boot, React, MySQL
Built a crowdsourced last-mile delivery application using Spring Boot (MVC architecture).
Applied Spring Design Patterns (Singleton, DAO, Adapter) and SOLID principles, improving system modularity.
Developed a React + REST API frontend enabling customers to request couriers and track deliveries.
Tech: OpenCV, SQLite
Built a face recognition-based attendance system using LBPH achieving 95% accuracy.
Developed a food-item recognition model using Haar Cascade with 84% accuracy for calorie estimation.
Analyzed BMI and calorie intake to evaluate public health outcomes.
📰 Research Paper:
"Tracking Impact of PM Poshan on Child’s Health"
Published in International Journal of Computer Engineering and Applications (IJCEA), 2023
Thanks for visiting my profile! Feel free to reach out or explore my projects. 🚀