Skip to content
View PETERMUTWIRI's full-sized avatar
💭
available
💭
available
5 followers · 5 following

Achievements

Achievement: Pull Shark

Achievements

Achievement: Pull Shark

Block or report PETERMUTWIRI

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.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

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

Report abuse
PETERMUTWIRI/README.md
PETERMUTWIRI

Hi there, 👋 I’m Peter Mutwiri

Full-Stack Engineer | AI-Agent Builder | CloudOps Engineer

Current mission: Ship production-grade AI copilots, keep the AWS bill low, and stop Stripe from eating SaaS margins.

---

🚀 Latest Fire (2024-2025)

Stack What Went Live Repo / Link
MutsynChub – Analytics Copilot Redis-driven event hub, DuckDB + pg-vector, 5 autonomous agents (embed, chat, KPI, entity, industry detector). 100 % TypeScript, serverless on Vercel. mutsynchub.app
Neon-Stack-Auth Refactor Fixed race-conditions, cache poisoning, TS strict-null, zero build errors. SSR role resolution, Google SSO, atomic logout hook.
Vibe-Code Agency 4 shipped sites in 30 days (Next.js, Tailwind, Stripe, Resend). 100 % prompt-to-prod, 0 % hand-written CRUD. Portfolio below
POS Plug-in Node bridge that pushes CSV sales → DuckDB lake nightly; used by 3 Kenyan retailers. petermutwiri/pos-sync

🧠 Agent Architecture I Use

  1. Ingest – S3 / Webhook → Redis Streams (event sourcing)
  2. Embed – OpenAI text-embedding-3 → pg-vector (DuckDB for OLAP)
  3. Plan – ReAct loop (function-calling) with deterministic guardrails
  4. Memory – Thread-level conversation memory, long-term Zep hybrid
  5. Ops – Vercel edge functions, CloudWatch alarms, SLO: 250 ms p95
  6. Wallet – Usage-metered → Stripe meter events → automatic top-up

☁️ Cloud & Ops

  • AWS – CDK-Terraform hybrid, VPC-less serverless, EventBridge pipes, S3 life-cycle 30 d → Glacier.
  • Oracle Cloud – Always-Free ARM VM running Plausible analytics + n8n workflows.
  • Vercel – All customer front-ends, Preview-per-PR, 0-downtime.
  • DuckDB-WASM – Browser-side OLAP, no backend for < 50 MB datasets.
  • Monitoring – Grafana Cloud (Prometheus), PagerDuty on 5 % error budget burn.

🛠️ Current Toolkit

TypeScript Next.js Tailwind Stripe OpenAI Redis DuckDB AWS CDK Terraform Docker GitHub Actions

📈 GitHub Pulse

Stats Top Langs

🤝 Open to

  • Fractional SRE / AI-agent gigs
  • Technical writing on serverless + AI patterns

Last commit: 2025-01-20 UTC

Pinned Loading

  1. caesars-cypher-app caesars-cypher-app Public

    This is an app used to decrypt and encrypt a massge

    Python 1

  2. Machine-Learning-Tutorials Machine-Learning-Tutorials Public

    Forked from eaedk/Machine-Learning-Tutorials

    This repository hosts some notebooks to guide people to start a Machine Learning project

    Jupyter Notebook 1

  3. pandas-revenue-calculation pandas-revenue-calculation Public

    Using panadas one of most useful modules to calculate revenue

    1

  4. Random-forest-model Random-forest-model Public

    Machine learning random forest model training and testing

    1

  5. Azubi-Africa/git-and-github-final-assignment-PETERMUTWIRI Azubi-Africa/git-and-github-final-assignment-PETERMUTWIRI Public

    git-and-github-final-assignment-PETERMUTWIRI created by GitHub Classroom

    Python

  6. Israel-Anaba/Indian-Start-up-Project Israel-Anaba/Indian-Start-up-Project Public

    As the data expert of your team, you have been tasked with investigating the Indian start-up ecosystem and recommending the optimal course of action for your team's venture into this space.

    Jupyter Notebook