Sreeshanth Ryali

AI & Robotics undergrad. Builder. Problem solver.

I'm a B.Tech student in AI & Robotics at SASTRA University, building at the intersection of language models, multi-agent systems, and applied AI. I care about systems that work — clean code, sharp reasoning, and things that solve real problems. When I'm not coding, I analyze financial markets and play chess.

GitHub LeetCode LinkedIn Lichess Chess.com Email
Sreeshanth Ryali

Projects

SentinelFinance

GitHub →
  • Designed a multi-agent financial reasoning system (Router, Researcher, Analyst) using LangGraph to process financial documents and generate personalized insights
  • Built a sandboxed Python computation engine to eliminate LLM arithmetic hallucinations
  • Per-user FAISS vector stores with real-time market data integration
  • Full-stack backend with FastAPI + MySQL, user auth, session-based chat memory

Python, FastAPI, LangGraph, FAISS, MySQL

Nexa | AI Assistant

GitHub →
  • ReAct-based AI agent orchestrating web search, Calendar, and Gmail via LangGraph workflows
  • Dual-LLM verification pipeline (generator + reviewer) to reduce hallucinations via iterative refinement
  • Interactive chat with reasoning trace visualization and tool usage transparency
  • Natural language orchestration of web search, Google Calendar, and Gmail actions

Python, FastAPI, LangChain, LangGraph

TaskFlow

GitHub →
  • Full-stack Kanban board for professors with drag-and-drop task management and session-based auth
  • Automated course planning by integrating LLMs to parse PDF syllabi and generate structured tasks
  • Custom probability distribution from real exam patterns, generated ~9,000 synthetic records
  • Trained a classifier to predict final grades from mid-term scores with noise injection

Python, Flask, MySQL, scikit-learn, LangChain

Experience

AI Intern — bEarly Technovations

Sep 2024 – Nov 2024
  • Built an automated gait analysis reporting system, reducing neurologist diagnosis time by ~60%
  • Engineered data preprocessing pipelines for multi-sensor gait time-series data
  • Applied unsupervised learning (PCA, clustering) to identify gait pattern anomalies correlated with early Parkinson's disease indicators

Technical Skills

Languages Python, Java, JavaScript
ML / AI TensorFlow, scikit-learn, Pandas, NumPy, LangGraph, LangChain
Backend FastAPI, Flask, Node.js, Express, MySQL, MongoDB
Tools Git, Linux, Docker, Postman, AWS

Achievements

Contact

Reach me at sreeshanthryali@gmail.com or find me on LinkedIn and GitHub.