Arnav Panigrahi
Software Developer, AI/ML Enthusiast
About Me
I'm a Computer Science graduate from the University of California, Riverside, with 2+ years of hands-on experience in software development and a strong interest in ethical, privacy-conscious AI.
I'm a software developer at AllCheer, where I build AI-driven web applications using React with Express.js and Flask backends. My work includes automating healthcare workflows with HIPAA-compliant systems, integrating GPT-powered APIs for intelligent document processing, and deploying real-time data pipelines with FastAPI, Supabase, and cloud-native infrastructure. I also develop internal tools for therapists and staff using platforms like Make and RESTful APIs, streamlining operations across the board.
I'm deeply interested in the intersection of Artificial Intelligence, Machine Learning, and Linux-based systems. My focus is on building responsible, real-world AI systems that align with open-source values and data privacy principles.
In my free time, I'm exploring NixOS and Deep Learning, with a growing interest in deploying ML models and experimenting with agentic AI architectures.
My Projects
Bedtime.ai - AI-Powered Interactive Bedtime Stories
Innovative React/TypeScript application that generates personalized bedtime stories using phi-3 fine-tuned model. Features drawing recognition with EfficientNet-B4, voice cloning with YourTTS, and a microservices architecture with FastAPI backend, PostgreSQL database, and Redis caching for optimal performance.
Rethink BH Automation – Real-Time Sync
Built a FastAPI backend that syncs appointment and authorization data from Rethink BH in real time. Emulated secure user login, deployed on Cloud Run with Docker and full observability stack, resulting in 90% cost savings and replacing a 24–48h latency pipeline.
RAG System with Pinecone & Groq
Architected a high-performance Retrieval-Augmented Generation system that processes documents in real-time. Integrated Pinecone vector database with Groq's inference API to achieve efficient responses and a 35% improvement in search relevance over traditional methods.
AI-Powered Media Analysis
Engineered a full-stack web application that analyzes online articles using NLP techniques to detect clickbait, assess content quality, and generate concise AI summaries. Achieved 92% accuracy in identifying misleading headlines and reduced reading time by 78% through AI summarization.
DeepArtDetect: AI vs Human Art Classifier
Developed a high-accuracy computer vision system that distinguishes between AI-generated and human-created artwork with 94% precision. Implemented and compared multiple deep learning architectures (ResNet, EfficientNet, Vision Transformer) to identify subtle patterns invisible to human observers.
HanziRecognize: Chinese Character Classifier
Built an interactive web application featuring real-time Chinese character recognition with 97% accuracy. Implemented a custom CNN architecture optimized for handwritten character recognition, with a responsive canvas interface for drawing and instant feedback.
My Skills
Programming Languages
- Python
- JavaScript
- TypeScript
- C++
- Java
Frontend Development
- React
- Angular
- Tailwind CSS
- Svelte
- Astro
Backend Development
- Node.js
- Express.js
- PostgreSQL
- Flask
Machine Learning
- PyTorch
- TensorFlow
- Neural Networks
- Computer Vision
- NLP & Transformers
Tools & Platforms
- Git
- Linux
- NixOS
- Docker
- Jupyter Notebook