5 Weekend AI Engineering Projects: Cost Routing, Caching, Evals & Observability

Bashiri Smith · Facebook reel · 2026-09-14 · 1:28 · 19,059 views · Open on Facebook

Topics: Portfolio Projects, LLMOps, Deployment & Monitoring, Evaluation (Evals) & Testing · Level: intermediate

Summary

Bashiri Smith suggests five AI projects you could build in a weekend. Each one deals with a real production problem: model cost routing, semantic caching, regression testing against a golden dataset, observability for multi-step pipelines, and documentation that updates itself through a GitHub Action. He says building them teaches more than any course. The video ends by inviting viewers to comment "5" to get a guide with build instructions and an AI engineer community.

Key points

• LLM cost autopilot: a routing layer in front of several LLM providers. It judges how complex each query is and sends it to the cheapest model that can still answer well.
• Semantic caching layer: middleware that spots requests that mean the same thing as earlier ones and returns the cached answer right away. This cuts latency and API costs.
• Model regression detector: a CI/CD-style pipeline that keeps testing LLM-powered features against a golden dataset. When a system prompt or model change lowers quality, it alerts the team on Slack.
• Failure forensics tool: an observability layer for multi-step AI pipelines. It traces every intermediate step, finds where a failure started, and flags the failure for evaluation.
• Self-healing technical docs: a GitHub Action that watches a codebase, notices code changes that affect the documentation, finds the outdated sections, and updates them.
• The creator's guide has build instructions for all five projects. The counts of extra projects don't match: the caption says 21 more, the spoken version says 10 more.

Resources mentioned

• Bashiri Smith's AI Projects Guide · pdf · drive.google.com · free
  The creator's guide to 27 AI projects, with full architecture, build steps phase by phase, interview talking points, and the stack for each project.
  Also in: 6-Step Framework for Building AI Projects Target Companies Care About (Bashiri Smith on Facebook · notes), Generate Job-Worthy AI Project Ideas from Recent Research Papers + ChatGPT (Bashiri Smith on Facebook · notes), 3 AI Portfolio Projects That Get Past the 6-Second Resume Scan (Bashiri Smith on Facebook · notes), 15 Production-Grade AI Projects to Go from Software Engineer to AI Engineer (Bashiri Smith on Facebook · notes) and 3 more
• BASWE.Ai Engineer (Skool community) · community · skool.com · paid
  The creator's paid community and program, with an AI learning roadmap (including the full ops and evaluation track), daily calls with engineers and recruiters, resume and portfolio help, and a job-search pipeline.
  Also in: Basic RAG Pipeline in 60 Seconds: From Documents to Grounded Answers (Bashiri Smith on Facebook · notes), Pointer to Bashiri Smith's Complete AI Engineer Roadmap for 2026 (Bashiri Smith on Facebook · notes), Step-by-Step Roadmap to a $200K+ AI Engineering Role (Bashiri Smith on Facebook · notes), How to Evaluate a RAG Pipeline: Retrieval vs. Generation (Interview Answer) (Bashiri Smith on Facebook · notes) and 76 more
• Slack · tool · slack.com · free · recommended by both Bashiri Smith & Melvin Vivas
  Team chat app where the user-feedback channel lives and where the agent replies to close the loop.
  Also in: Dots Demo: A Voice AI Agent Handling Travel Booking and Feedback Triage (Melvin Vivas on X · notes), Grok Team Bots: Shared AI Teammates in Slack and Grok (Melvin Vivas on X · notes), ChatGPT Voice Update: Plugins, Model Switching and ChatGPT Work (Melvin Vivas on X · notes), Devin Gets a Mac VM: AI Agent Builds and Ships iOS Apps (Melvin Vivas on X · notes) and 4 more
• GitHub Actions · tool · github.com · free · recommended by both Bashiri Smith & Melvin Vivas
  GitHub's CI/CD automation platform, used to run the self-healing docs workflow.
  Also in: Use Claude Code /loop to Watch and Auto-Fix Failing GitHub Actions (Melvin Vivas on X · notes)

Try this

• Pick one of the five projects and build it over a weekend.
• Comment "5" on the video to get the guide link and the community invite.
• LLM cost autopilot: a router that scores query complexity and sends each query to the cheapest model that can still answer well
• Semantic caching middleware for LLM APIs that serves semantically similar requests from a cache
• Model regression detector: a CI/CD pipeline that tests LLM outputs against a golden dataset and sends Slack alerts
• Failure forensics / observability tool that traces multi-step AI pipelines and finds where failures start
• Self-healing technical docs: a GitHub Action that finds outdated docs after code changes and updates them