That TDD makes the top of the list of the used skills is a good sign. It means the agent is taking my instructions seriously! #AIEngineering #TDD #CLIFirst #AIFirst
@geekcoder.bsky.social
@geekcoder.bsky.socialThat TDD makes the top of the list of the used skills is a good sign. It means the agent is taking my instructions seriously! #AIEngineering #TDD #CLIFirst #AIFirst
@gosuraj.bsky.socialReducing tokens with Presigen AI #ai #promptengineering #aiengineering #tokenoptimization #indiedev www.youtube.com/watch?v=kbU-...
AWS just released a guide on migrating multi-model AI agents to Amazon Bedrock AgentCore runtime. This shift reduces operational overhead by offloading container orchestration, scaling, and observability to managed services. Developers can now focus on agent logic, not infrastructure. #AIEngineering
@yurekilab-en.bsky.socialGave every task my coding agent runs a hard dollar cap: $1.50. Hit the cap and it stops, writes down what's done and what's missing, and exits. No heroics. A month in, the tasks that blew the cap were rarely hard. They were underspecified by me. #claudecode #ai #aiengineering #buildinpublic
@yurekilab-en.bsky.socialMy coding agent shipped a page that ran 412 DB queries per load. Every test passed. The tests seeded 3 rows. Production had 400. Now my seed script makes 1,000 of everything, and the agent pastes the query count in its summary. #claudecode #ai #aiengineering #buildinpublic
@loopandretry.bsky.socialContext engineering: what fits and what gets dropped
"Context engineering" is getting used as a rebrand for prompt writing, and that undersells it into uselessn…
https://loopandretry.github.io/posts/context-engineering-what-fits-what-gets-dropped/?ref=bluesky
@yurekilab-en.bsky.socialMy coding agent has never seen a single API key. Secrets sit in one file it's told not to read. A tiny lookup command returns a value only at runtime, and every call is logged. Two months in: zero keys in commits or transcripts. ~20 lines of shell. #claudecode #ai #aiengineering #security
Scaling agentic AI on AWS isn't just about infrastructure—it's about handling non-deterministic compute, IAM scoping, and token costs. Here's what we learned from real-world production setups. #AIEngineering #AWS #AgentDevelopment
@octalchip.bsky.socialAI agents need access to more than just a model.
MCP provides an open standard for connecting AI applications with tools and data — including APIs, databases, files and business systems.
Build connected AI agents with @octalchip.
#MCP #AIAgents #AIEngineering #GenerativeAI
@tbone.se#MCP #AIEngineering #MVPBuzz #CommunityRocks
Made with AI · Human-reviewed
@upsun.comWant to know how Upsun Dispatch runs workflows? It has three big primitives: workflow, sandbox, and integrations. Workflows are the core of our product. Let's dive into the details 👉 upsun.com/blog/how-ups...
#SDLC #AIEngineering #EngineeringLeadership #Upsun #UpsunDispatch
@jackceoai.bsky.socialI don't manage people. I manage systems. Systems compound. One dropped relay collapses the whole race. Design transitions first. #AIEngineering #BuildInPublic
@hackernoon.comThe EU Edge AI Workshop highlights Europe’s push for distributed compute, trusted infrastructure, and 10,000 edge nodes by 2030. #aiengineering
@temprhq.bsky.socialDefine the minimum context window and capabilities your workflow needs before replacing a model. TemprHQ helps compare candidates against that baseline. #AIEngineering #LLM https://temprhq.io/models
@yurekilab-en.bsky.socialCaught my coding agent adding a package for something the standard library already does. New rule in its instructions: name the stdlib option first and say why it falls short. About half the time, the package never gets added. #claudecode #ai #aiengineering #buildinpublic
@qconferences.comShuangfeng is speaking at QCon AI New York, Dec 15-16.
@heyymeraki.bsky.socialIf you have ever wondered what happens when you send a message to an LLM, this goes inside the API call, what you are sending every time, the controls you have over what comes back, and what is inside the response.
@groxio.bsky.socialToday 12pm EDT
Groxio Live with @redrapids.bsky.social
Where does MCP fit in an LLM workflow?
Bruce will demo an LLM harness, show MCP in context, and answer questions live.
Thursday, September 17 at noon ET / 16:00 UTC. Free: grox.io/live
@sayedev.bsky.socialOne thing is becoming increasingly clear:
The model itself is no longer the product.
The system around it, tools, memory, orchestration, and the execution loop are becoming just as important as the model itself.
#AI #AIEngineering #AgenticAI #AI #MachineLearning #AIEngineering #Inference
@yurekilab-en.bsky.socialBuilt a tiny remote panel to check on my AI agent from my phone. Planned for logs, progress bars, token counts. What I actually use: the pending diff and two buttons — approve or reject. Everything else was pretend supervision. #claudecode #ai #aiengineering #buildinpublic