Sanyam Garg
Age:
Madison, WI
I am a passionate UW–Madison junior pursuing a B.Sc. in Computer Science & Data Science. I thrive on solving complex problems and exploring innovative technologies.
Education
University of Wisconsin–Madison
Aug 2023 - May 2027
Degrees
Relevant Coursework
Tools
Programming Languages
Python, Java, C, Dart, SQL, Bash
DevOps & Infrastructure
Docker, Tailscale, Linux, Cloudflared, Git, Conda, Snakemake
Machine Learning & Data Science
PyTorch, NumPy, Pandas, Deep Learning, NLP
Web & Application Development
Flask, Flutter, Firebase, Node.js, REST APIs, WebSocket
Experiences
Century Financial
Dubai, UAE | June 2025 - Aug 2025
AI & Backend Development Intern
- AI Trading Assistant: Engineered the Python backend, integrating OpenAI's API with a RAG system to fetch live user portfolio data and generate personalized financial insights.
- Implemented the Model Context Protocol (MCP) to orchestrate tool calls, enabling the AI to securely manipulate data and store context across sessions while optimizing token usage.
- Real-Time Co-Browsing Engine: Built using Node.js and WebSockets. Utilized DOM mirroring instead of video streaming to drastically reduce bandwidth and latency.
- Developed a client-side masking feature to redact sensitive financial data before network transmission. This solution replaced a third-party tool and saved the firm $3,000+ annually.
Keles Group (UW–Madison)
Madison, WI | May 2025 - Present
Deep Learning Research Assistant
- Developing and evaluating deep learning models (VERMIN framework) in PyTorch to accurately predict DNA methylation from large-scale human/mouse genomic sequences.
- Performed extensive Exploratory Data Analysis on a massive 22-million-row dataset using Pandas in JupyterLab to investigate genetic variation and epigenetic markers.
- Implemented data processing workflows (Snakemake) on an HPC cluster and generated data visualizations to validate pipeline output.
Projects
AirPipe – CLI File Transfer Tool
GitHubBuilt a CLI tool in Go that transfers files between servers and devices by scanning a QR code. No apps, no accounts. Just scan and download.
Architected a zero-knowledge relay using WebSocket where encryption keys never leave the client. Uses NaCl secretbox (XSalsa20-Poly1305) encryption with the key embedded in the URL fragment, so the relay only ever sees encrypted bytes. Open sourced with 50+ GitHub stars, featured on r/selfhosted.
Private Cloud & Server
Network MapWanted full control over my data and a playground to experiment with infrastructure. Built a hybrid-cloud setup spanning cloud VPS and bare-metal Linux, connected securely with a Tailscale mesh network.
Now running 20+ services in Docker containers, from media automation to development tools. Used Cloudflare Tunnels for public access without exposing my IP. Wrote custom monitoring in Python and automated daily Bash/Cron backups to an off-site location.
After exposing my server's SSH to the internet, I got curious about who's actually trying to get in Within minutes, bots worldwide were hammering it with credentials like "root:admin". Built this to watch it happen in real-time.
Created a Python (FastAPI) pipeline that tails system logs over SSH, extracts attacker IPs, resolves their locations, and serves everything through a REST API that the dashboard polls every 1.5 seconds. Deployed with Docker and Cloudflare Tunnels. The map uses Leaflet.js to animate each attack as it happens.