Praneeth Ravuri’s developer project resources

Using these resources

Praneeth Ravuri’s public project repositories are the source for setup instructions, implementation details, dependencies, and licensing. Follow each repository’s current README before running a project. This portfolio provides a directory and descriptions; it does not host an API, authentication service, or MCP server. Pitstop is an MCP project whose installation and configuration belong to its repository.

Tether

Coding agents in separate terminals need a way to talk and coordinate edits. Tether gives them a shared inbox through a local CLI. Agents can send questions, wait for replies, leave handoffs, and claim files they plan to work on. A Go daemon stores messages in SQLite and connects sessions across Git worktrees. Technologies: Go, SQLite, Unix Sockets.

• Tether source code
• Tether README and setup

Gary

Gary takes a master resume and a job description and produces a tailored Word document. Three agent passes analyze the role, draft the resume, and check the result against the source material. A document template keeps the formatting consistent, and Google Sheets records the application details. Technologies: Python, CrewAI, Pydantic, docxtpl, Google Sheets API.

• Gary source code
• Gary README and setup

Pitstop

Pitstop brings Formula 1 data into an AI chat through MCP. Its tools cover race results, standings, schedules, telemetry, and news. For questions about a race, it can compare lap pace, summarize stints, and calculate how lap times change over a tire stint. It combines FastF1, Jolpica, and OpenF1 with a queryable SQLite database. Technologies: Python, FastMCP, FastF1, HTTPX, SQLite.

• Pitstop source code
• Pitstop README and setup

Smart Traffic

A university team project exploring how reinforcement learning can control a four-way intersection. A SARSA agent chooses which direction gets a green light, using accumulated vehicle delays to represent traffic conditions. Pygame shows the traffic simulation, while learning curves track the agent’s behavior during training. Technologies: Python, NumPy, Pygame, Matplotlib.

• Smart Traffic source code
• Smart Traffic README and setup

Notstuck

Upload a PDF, Word document, or text file, then ask questions about it. Notstuck splits the document into chunks, embeds them, and retrieves relevant passages from Pinecone. A chat interface shows the answer alongside document references. The assistant also has a web-search tool for questions beyond the uploaded files. Technologies: Next.js, TypeScript, FastAPI, Pinecone, CrewAI, OpenAI Embeddings.

• Notstuck source code
• Notstuck README and setup