Computer Science

Hi, I'm Riyam Patel

AIML, Data Science, Cloud and Big Data, Information Retrieval. From building multi-agent RAG systems to designing cloud-native platforms, I create scalable solutions.

Experience

Teaching Assistant – Machine Learning
New York University
Jan 2025 – Present
  • Guide 50+ graduate students through ML fundamentals including preprocessing pipelines, supervised/unsupervised learning, deep learning, and model optimization techniques
  • Conduct weekly office hours to debug Python code, explain algorithms, and assist with PyTorch implementations
  • Develop comprehensive learning materials covering neural networks, transformers, and advanced ML concepts
AI Intern
Shepherd Ventures
Jun 2025 – Aug 2025
  • Developed ML pipeline for portfolio optimization using Mean-Variance Optimization frameworks to identify and rank low-volatility stocks across 500+ securities
  • Built and trained supervised ML models in Python to forecast stock volatility and returns, achieving 15% improvement in risk-adjusted portfolio performance
  • Conducted statistical backtesting of trading strategies using historical market data, collaborating with senior investors to validate model predictions

Featured Projects

Multi-Agent RAG System
2025

Engineered a sophisticated multi-agent RAG system with specialized QueryAnalyzer, Retriever, Generator, and Verifier agents. Achieved 67% improvement over baseline through dense embeddings and query reformulation. Built ETL pipelines integrating MongoDB and Qdrant vector database, enabling sub-second query response times.

Python MongoDB Qdrant ClearML Gradio
Multi-Tenant DBaaS
2025

Architected a cloud-native multi-tenant DBaaS with automated provisioning and parallel migration engine. Orchestrator coordinates 100 workers via SQS queues for concurrent tenant migration with retry mechanisms. Implemented Master-Replica database replication for high availability.

AWS Lambda Redis API Gateway DynamoDB SQS
Query-Based Audio Retrieval
2025

Designed text-to-audio retrieval system using CRNN and TCN architectures with custom contrastive learning loss functions (InfoNCE, VICReg, Cosine). Achieved 85% retrieval accuracy across multi-domain audio datasets, outperforming baseline models by 22%.

PyTorch CRNN TCN Deep Learning
Multi-Threaded Web Crawler
2025

Built scalable web crawler supporting 20-100 concurrent threads with intelligent priority-based URL queue management. Implemented HTTP connection pooling and atomic thread-safe operations, reducing request latency by 40-60% while maintaining robots.txt compliance across 10,000+ crawled pages.

Python Multi-threading HTTP Pooling

Technical Skills

Languages & Core

Python SQL C/C++ Java

ML & AI Frameworks

TensorFlow PyTorch Hugging Face Scikit-learn Neural Networks Transformers Computer Vision NLP

Cloud & Infrastructure

AWS Docker CI/CD MongoDB PostgreSQL Redis Hadoop Spark

Specializations

Distributed Systems Machine Learning Data Science Information Retrieval ETL Pipelines Cloud Computing

Get In Touch

I'm currently seeking software engineering and machine learning opportunities. Whether you want to discuss distributed systems, ML infrastructure, or potential collaborations, I'd love to hear from you.