Razam Sherwani

CS Student & Software Engineer

Hi, I'm Razam, a Computer Science student at Georgia Tech concentrating in the Intelligence and Theory threads. My work bridges rigorous mathematics and practical engineering, spanning from algorithm research to building the backend infrastructure for GT's largest AI tournaments. I built this site to showcase my previous projects and to serve as a home for my blog, where I explore the algorithmic and mathematical topics I find most interesting.

Currently

Software Engineer
ByteFight (Georgia Tech)
Undergraduate Teaching Assistant
CS 2050: Discrete Mathematics · Georgia Institute of Technology
Undergraduate Research Assistant
Algorithmic Graph Theory · Georgia Institute of Technology

Education

Georgia Institute of Technology
Bachelor of Science, Computer Science — Concentrations: Intelligence & Theory · GPA 4.0Atlanta, GA
  • Relevant coursework: Design & Analysis of Algorithms, Data Structures, Intro to AI, Perception & Robotics

Industry Experience

Software Engineer – Present
ByteFight (Georgia Tech) · Java, Spring Boot, Python, PostgreSQL, RabbitMQ, GitHub ActionsAtlanta, GA
  • Closed a live unauthenticated API leak on the bot-vs-bot AI competition platform behind Georgia Tech courses: match replays were handing 200 student teams the instructor bots' strategies and hidden game state.
  • Shipped rate-limited practice matches against benchmark bots: per-team quotas enforced by one atomic PostgreSQL upsert doubling as audit log, and practice traffic split from graded matches by RabbitMQ routing key.
  • Moved graded bot evaluation server-side after proving across 10 candidate architectures that obfuscation cannot protect agents on student machines; hardened the GitHub Actions pipeline compiling bots for 4 platforms.
Backend Founding Engineer Intern –
Lume Security · Rust, Python, Azure AKS, PostgreSQL, Redis, Prometheus, GrafanaAtlanta, GA
  • Delivered a Microsoft Teams bot service as project lead, from empty repository to production in under 2 months across ~20 merged PRs covering authentication, sessions, LLM agent replies, and Kubernetes deployment.
  • Ran the company's first capacity characterization of its Rust workflow engine: sustained 4,000 req/s and proved the bottleneck was a 10-connection PostgreSQL pool saturating 10x earlier in the cloud than documented.
  • Reproduced a silent production failure 3 times on demand, showing an 800x buffer-sizing error rejected 100% of writes at 5% of capacity as health checks returned 200; shipped Prometheus/Grafana alerting to detect it.
  • Automated joiner-mover-leaver (JML) onboarding and offboarding for 2 enterprise tenants as the first hands-on author and critic of the company's custom workflow engine, filing 116 defect and platform-gap tickets.
VR Security Engineer (CAPTCHA Replay) –
Georgia Tech VR Club · Python, ChromaDB, LSH, HNSWAtlanta, GA
  • Engineered a biometric replay detection system for VR using Locality-Sensitive Hashing (LSH) and Hierarchical Navigable Small Worlds (HNSW) to identify spoofed motion sequences.
  • Built a vector search pipeline ingesting 9-channel spatial telemetry (head/hand tracking) into 5-second temporal windows to detect adversarial variations like cropping and noise injection.

Academic & Research Experience

Undergraduate Teaching Assistant (CS 2050: Discrete Mathematics) – Present
Georgia Institute of TechnologyAtlanta, GA
  • Instructed students in foundational theoretical computer science, emphasizing concepts critical to AI and systems engineering: Big-O complexity, finite probability, and cryptographic number theory (RSA).
  • Evaluated mathematical proofs and problem sets, coaching students on formal logic, strong induction, and the probabilistic method to develop rigorous algorithmic problem-solving skills.
Undergraduate Research Assistant (Algorithmic Graph Theory) – Present
Georgia Institute of Technology · Python, LaTeXAtlanta, GA
  • Investigated polynomial methods for spanning tree enumeration, utilizing the Deletion-Contraction principle to mathematically map Tree Weight Polynomial coefficients directly to Minimum Spanning Tree frequencies.
  • Engineered a Python backtracking solver to empirically test theoretical graph recurrences, designing a custom reversible Union-Find data structure that omits path compression to enable O(1) state rollbacks.

Technical Skills

Languages
Rust, Python, Java, C++, C#, JavaScript, SQL, LaTeX
Frameworks
React, FastAPI, Spring Boot, Node.js, Django, Flask, Celery
Cloud & DevOps
Azure, Kubernetes, Docker, Helm, GitHub Actions, Vault
Data & Observability
PostgreSQL, Redis, RabbitMQ, Prometheus, Grafana, ChromaDB
Libraries
TensorFlow, Scikit-learn, Pandas, NumPy, Matplotlib