Skip to content
View Vectorplayzz's full-sized avatar

Block or report Vectorplayzz

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. 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
Vectorplayzz/README.md

VECTORPLAYZZ. AI engineer: agents and infrastructure. I build AI agents, and the infrastructure that keeps them honest. Now: fine-tuning LLMs, building agents that ask before acting, running my own servers.

PortfolioLinkedIn

About

Agents that ask first: evidence before conclusions, and a human approval before anything that can't be undone. Ship the whole thing: not just the model, but the API, the interface, auth and the tests around it. Run what I build: every live demo runs on servers I set up, harden and keep alive myself.

Featured work

SITREP: on-call agent that investigates incidents end to end, proves the root cause with statistics, and stops for a human before touching prod. 77 seconds to root cause, 4 approval gates. Python, MCP, SciPy.OpsLoop: SRE agent that checks health actually held after a fix, and rolls back on its own when it didn't. 98 tests, injection-hardened logs. Python, Discord, Loki. Cognify: learning platform that turns any topic into lessons, quizzes and coding practice, with self-hosted RAG and a sandboxed code runner. 94 roadmaps, 7 sandboxed languages. TypeScript, React, Ollama.ClutchR: esports tournament platform with teams, matches, leaderboards and a wallet, behind JWT auth, role-based access and rate limits. TypeScript, Express, Prisma.

Play Cognify livePlay ClutchR livePlay MiniMart live
Also here: MiniMart, multi-branch inventory with demand forecasting · CAPTCHA Pro, the world's most frustrating CAPTCHA (on purpose)

Toolbox

Toolbox. AI and ML: Python, PyTorch, Hugging Face, pandas, SciPy. Agents: MCP, Ollama, RAG, approval gates. Web: TypeScript, React, Node.js, Express, Prisma, MySQL. Infra: Linux, Docker, NGINX, systemd and PM2. Cloud and CI: Google Cloud, Cloudflare, GitHub Actions.

Pinned Loading

  1. sitrep-incident-commander sitrep-incident-commander Public

    Autonomous incident commander: investigates production incidents with statistics, delegates to subagents, and stops for human approval before touching prod. Built on TrueForge.

    Python

  2. devops-failure-recovery-agent devops-failure-recovery-agent Public

    OpsLoop: chat-driven SRE agent that diagnoses from cited evidence, gates fixes on human approval, then verifies recovery held and rolls back if not.

    Python

  3. cognify cognify Public

    AI learning platform: any topic becomes lessons, quizzes and coding practice with an AI tutor. Self-hosted RAG, sandboxed code runner, web search.

    TypeScript

  4. clutchr clutchr Public

    Esports tournament platform: teams, matches, leaderboards and a wallet, with JWT auth, role-based access and rate limiting.

    TypeScript

  5. minimart minimart Public

    Multi-branch retail inventory system: stock transfers, role-based views, reports and demand forecasting.

    JavaScript

  6. worst-captcha worst-captcha Public

    CAPTCHA Pro: the world's most frustrating verification. A joke, and a tour of the dark patterns real products use.

    JavaScript