Skip to content
View VirajMishra1's full-sized avatar

Block or report VirajMishra1

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
VirajMishra1/README.md

Viraj Mishra

CS + Math @ Penn State, Schreyer Honors College. Graduating December 2026.

ICPC team @ Penn State. PennApps winner. HackPSU winner. Currently researching LLM agent security and RAG for political science codebook annotation.

open to work: full time roles starting January 2027. open to relocating, on F-1 with CPT/OPT for now.

interested to work anywhere where the work is hard and shipped fast.

linkedin / x / viraj.mishra.81@gmail.com

Python C C++ TypeScript JavaScript Verilog Next.js React Tailwind Three.js FastAPI Streamlit OpenAI PyTorch HuggingFace scikit-learn Gymnasium NumPy LangChain LangGraph FAISS OpenCV MediaPipe SQL Flask Pandas Azure Docker Git SciPy GitHub Actions Vercel Cloudflare

most of what's below is live. click around.

📈 polymath · live · demo
Bloomberg terminal for prediction markets, built at CMU TartanHacks. live orderbooks, arbitrage scanner, Monte Carlo sims, portfolio VaR.
🧳 wander agent · demo
MCP server that turns Claude into a travel agent with memory. 66 tools, zero API keys. pip install wander-agent
🩺 medtrace · demo
clinical decision support over 15,000 PMC-Patients records. hybrid BM25 + FAISS retrieval, 13-stage pipeline. Palantir build challenge.
📦 loadout · demo
the npm for AI coding agents, built at OpenAI Build Week. 12 agents, 241-repo catalog screened daily. npm install -g loadout-ai
worldcup-forecaster · dashboard
Dixon-Coles Poisson model, fit on 13,779 matches. 71% W/D/L on 80 locked predictions, 78% in the knockouts.
🔭 bench · live
observability for AI agents. wrap in one call: dashboard, LLM evals, failure clustering, README badge.
🦖 dino trainer · demo
double DQN agent for Chrome dino. curriculum learning, behavior cloning warm start. avg score 167.95 — better than me.
🎳 jarvis air bowl · live · demo
bowling you play by swinging your arm at your webcam. MediaPipe swing tracking, Cannon.js physics.
🟩 wordle signal · demo
Chrome extension that solves Wordle with Shannon entropy. also solves Sedecordle (16 boards at once).
🚨 crisgo · live · demo
crisis navigation, Columbia hackathon. Dijkstra routing with 600m penalty zones, AR camera nav.
📚 studybuddy
upload a pdf, ask questions or generate a quiz. won best Streamlit + AI at HackPSU.
🌱 sustain-a-prompt · live
paste a prompt, see energy/CO2/water cost, get a leaner version back. cuts tokens ~40%. won MLH Best Auth0 at PennApps.

wc 2026, according to my model

dixon-coles poisson model, fit on 13,779 matches. predictions locked to git before kickoff. full table + every scoreline

🏆 spain won. model called the final correctly: 🇪🇸 spain 62% vs 🇦🇷 argentina 38%, predicted 1-0, result 1-0.

71% W/D/L accuracy on 80 pre-kickoff locked predictions (57/80, random guessing gets 33.3%). 78% in the knockouts (25/32). time-decay weighting, Transfermarkt squad-value corrections, xG substitution for WC 2022.

Image

experience + research

🏫 Penn State - undergraduate research assistant, Jan 2026 to present

Dr. Qunhua Li & Dr. Cassandra Tai

  • Example-RAG classifies political text by policy using retrieved labeled examples; no training needed
  • beats Halterman & Keith LoRA (F1 0.58 vs. 0.547), zero training; found 28% scoring error in their benchmark
  • co-authoring paper on Example-RAG with Dr. Li & Dr. Tai

Dr. Suman Saha — honors thesis

  • benchmarking prompt injection attacks on LLM agents using InjecAgent across 17 attack tools and 197 test cases
  • developing a dual-agent isolation architecture that sandboxes untrusted input from tool execution

teaching assistant, August 2025 to present

  • leading labs for 380+ students in digital logic design: Boolean algebra, FSMs, Verilog HDL

💼 SIG Group - software and AI engineer intern, June 2025 to August 2025

  • built a scheduling platform with React and Flask replacing Excel, reducing planning time from 3 hours to 30 minutes
  • deployed a RAG pipeline (Azure Search) over internal docs, enabling 130 engineers to resolve queries instantly
  • fine tuned YOLOv8 for defect detection on Azure ML, hitting 84% precision with only 100 training samples

classwork

the course projects live in private repos, but ask me about any of these:

  • ⚙️ 5-stage pipelined MIPS CPU - 32-bit RISC processor in Verilog. full 5-stage pipeline with custom hazard detection and forwarding units so data hazards don't stall execution. CMPEN 331 final
  • 🧠 dynamic memory allocator - malloc, free and realloc from scratch in C. implicit free list with boundary tags, then an explicit LIFO free list so find-fit only scans free blocks, then footer optimization with a prev_alloc bit packed into headers so coalescing works without footers. CMPSC 473
  • 🗺️ x86-64 kernel virtual memory - built a working kernel in C: full 4-level page table hierarchy (PML4E through PTE) in raw memory, identity-mapped 4GB for the kernel, separate user-space mapping, loaded CR3 myself, syscall dispatch from scratch. CMPSC 473
  • 🔀 Go-style buffered channels in C - blocking and non-blocking send/receive plus a channel_select that multiplexes n channels at once, built on Pthreads mutexes, condition variables and a shared semaphore. CMPSC 473
  • 💬 mycord - multithreaded terminal messenger in C. concurrent Pthreads server, non-blocking TUI straight from termios and ioctl. CMPSC 311 final
  • 📐 conformal prediction for regression - ICP, CQR, jackknife and jackknife+ from scratch in Python. all four hit the 90% coverage target on heteroscedastic insurance data with zero distributional assumptions, CQR's adaptive intervals beat ICP's fixed-width bands, validated against MAPIE. MATH 415 honors project. colab

right now

working on research and answering my email.

if you're hiring new grads for jan 2027 fastest way to reach me is viraj.mishra.81@gmail.com. i reply within a day.

snake eating my contribution graph

Pinned Loading

  1. wander-agent wander-agent Public

    Your AI's personal travel agent — 66 tools, zero API keys. Flights (incl. fares airlines hide), hotels, visas, weather, points & miles, via MCP.

    Python 7 1

  2. polymath polymath Public

    bloomberg terminal for prediction markets

    TypeScript

  3. jarvis-air-bowl jarvis-air-bowl Public

    bowling you play by swinging your arm at your webcam

    JavaScript 1 1

  4. wordle-signal wordle-signal Public

    solve both wordle or sedecordle using entropy

    JavaScript

  5. medtrace medtrace Public

    clinical decision support tool I built for the Palantir build challenge

    TypeScript

  6. loadout loadout Public

    The package manager for AI coding agent extensions. Discover, install, update, optimize, and roll back skills and MCP tools across Codex, Claude Code, Cursor, and more.

    TypeScript 3