Prathu Vijayvargiya
SDE Intern @ Interact AI
B.Tech CSE, SRM Institute of Science and Technology (2022 - 2026)
Software Engineer with startup experience in TypeScript, Python, AWS and AI/ML technologies. I build and deploy backend services, AI Agents, and scalable cloud infrastructure. Currently working on AI-powered Interactive Demo products at Interact AI in Bangalore.
I'm passionate about turning ideas into real-world systems, from medical image classifiers and sign language translators to developer tools and real-time audio processing platforms.
Experience
- Architected backend services using Python and Node.js for 2 flagship AI products, owning end-to-end development from system design to production deployment.
- Implemented RAG-based optimization engines and advanced prompt engineering strategies, improving user engagement and reducing model hallucination through robust guardrails.
- Designed and deployed scalable AWS infrastructure (EC2, S3, Lambda, SES), reducing infrastructure costs and handling high volumes of daily API requests.
- Deployed JWT-based authentication with RBAC and OTP-based signup across multiple microservices, securing access for enterprise clients.
Skills
| Languages | Python, SQL, TypeScript, JavaScript, Bash |
| ML / AI | PyTorch, scikit-learn, NumPy, Pandas, LangChain, Hugging Face |
| Backend | FastAPI, Node.js, Express.js, Django |
| Databases | MySQL, PostgreSQL, MongoDB, Redis, FAISS, Pinecone |
| DevOps / Cloud | AWS (EC2, S3, Lambda, SES, RDS), Docker, GitHub Actions, ELK Stack |
| Tools | Git, Google Colab, Power BI, Postman, Linux |
Projects
Medical Image Classifier
End-to-end deep learning system using fine-tuned ResNet-50 to classify medical images across X-ray, CT, MRI, and ultrasound modalities. Achieved 93% accuracy with stratified cross-validation. Deployed as a Flask REST API and Streamlit web app.
TokenOptEnv
An OpenEnv environment for CAMRE: a cost-aware meta-reasoning benchmark for code and log tasks. Agents learn how to manage context, retrieval, memory and tool use under real cost constraints.
Spectrolingua
Comprehensive audio processing studio with a Django API and Streamlit frontend. Supports audio format conversion (9+ formats), waveform/spectrogram visualization, and speech transcription across 35+ languages.
AI-Powered Log Analysis System
Real-time API monitoring system using Isolation Forest and Prophet time-series forecasting for anomaly detection. Containerized with Docker and ELK stack, processing 500+ logs/min with sub-5s latency. 87% detection accuracy.
Continuous Sign Language Translator
Real-time ASL interpreter achieving 95% accuracy using multimodal deep learning (3D-CNN, LSTM, attention mechanisms, transformer decoder). Optimized inference time by 40% through quantization and hardware acceleration for sub-200ms translation.
LeetSync
Automated LeetCode-to-GitHub sync tool. Syncs accepted submissions every 6 hours via GitHub Actions and LeetCode GraphQL API. Organizes solutions by difficulty and language with multi-language support.
Certifications
Education
SRM Institute of Science and Technology, Chennai
Relevant Coursework
- Machine Learning
- Data Science
- Artificial Intelligence
- Data Structures & Algorithms
- Database Management Systems
- Software Engineering
- Object Oriented Programming
- Cloud & DevOps Essentials