Building Morrow: A New Project and My Plan for What's Next
From today, I'll start working on a new open-source project called Morrow. It's inspired by Grok, OpenAI Bots, and similar personal AI assistants.
I'm not building this just to have another portfolio project on my GitHub. I genuinely want to build something that I can use in my day-to-day life and eventually turn into a proper product.
Here's what I'm thinking:
I need to find a better job before May 2027, and I hope Morrow will create a lot of opportunities for me. It's a chance to showcase a wide range of engineering skills through something real, rather than just working on small, disconnected projects.
I also want to start a separate system programming project that I'll work on every weekend. I'll record YouTube videos documenting the process, explaining the concepts, and doing Rust code-alongs.
You might think I'm taking the long route or doing things differently, but I genuinely believe this is the approach I should take.
I don't think I'll get many opportunities just by sending out hundreds of applications. I'll still apply, obviously, but I also want to do everything within my control to increase my reach and visibility. Building real systems, contributing to open source, sharing what I learn, and putting my work out there.
A little more about Morrow:
You'll be able to use it with your existing AI subscriptions or your own API keys.
I'm following a pricing philosophy similar to Amp.
It'll be open source, and you'll be able to self-host it.
I'm not going to vibe-code this (except for the landing page). I'll carefully build it, review the code, think through the architecture, work on evals, and make sure the engineering is solid.
I don't want this to be just another toy project. I want to build something I can depend on and something that other people might find useful too.
I'll share updates every single day.
Let's see where this goes. Peace out βπΌ
Finished 6th video from this Stanford's "Engineering AI Agents" course.
It was all about understanding all the different AI Agent Design Patterns like React, CodeAct, Reflexion, Workflow vs Agent.
This video also touched on the topic of context engineering and Pi harness too
I'm actually very excited about the next one as that is completely on Agent Memory and that is super interesting to me as I'm planning to build my own VectorDB as well to learn how this vector DB concept works
Finished 5th video from this Stanford's "Engineering AI Agents" course.
It was the most hands on lesson yet, learned about different frameworks like DSPy, Langraph, how to make agents durable using Temporal and observability
The best thing about this course is this course teaches us how to engineer AI Agents and that means its about about building AI Agents instead its about adding the other surrounding stuff that make our agents reliable, observable, durable, etc and this is exactly what I wanted to learn as there are 100s of courses regarding building an agent but barely any resource that teaches about concepts that surround our agents
@grad_ai assignments are posted on their website I'm not doing them because I'm currently building a cloud coding agent as a project and applying concepts there for example durability and observability from this lesson is very much my next target for my cloud coding agent project
Finished 4th video from this Stanford's "Engineering AI Agents" course.
It was the most in depth tool calling lecture I've ever seen. It explains from the very beginning of what are tools, how LLMs learn to call tools and a bunch of concepts related to tool calling
I just like how each lecture explains the concept like a evolution of the concept from how ti started to where we are at right now
Finished 3rd video from this Stanford's "Engineering AI Agents" course.
It was completely about RAG. The thing that I'm liking about this course is its not hiding mathematical equations and papers that are involved in these concepts
Next up tool calling, super excited to learn about that
@grad_ai Yes there are questions in between sections but I would say those questions and the answers are very valuable and also it's like a total of 5-10 minutes out of around 70 minutes of session so it's not that big of a deal for me
From next month, I'll start looking at vector DB and how to build my own.
If you are reading this post and can share any resource that can guide me be it a research paper, book or video or whatever. I'm ready to learn from that
Please share any resource you can
Finished 2nd video from this Stanford's "Engineering AI Agents" course.
It was mostly covering a ton of things around LLMs, training, serving and from the 3rd video it'll start picking up more towards AI Agents. Really excited for upcoming videos
Day 5 of Inference Engineering
I finished reading the 2nd chapter from "Zero to SGLang", its an intro to some of the core concepts of inference like prefill, decode, rooflining and this was a math heavy topic but as I was already familiar with these concepts I completely very quickly, it was kind of a revision of these concepts for me
@SunnyKu95684464 yeah I guess they just want to cover all the bases as quickly as possible to start focusing on AI agent topic, will go through the 3rd one tomorrow
I've started learning about AI Agents from this absolutely awesome course from Stanford "Engineering AI Agents"
I've always loved courses that actually cite papers instead of just explaining the concepts and move on. I
m so excited for this one because there are tons of papers to read and learn about this exiting field.
I've finished the 1st video, onto the 2nd one
You can find my notes on my notes url ,that is in my bio
@SunnyKu95684464 yes I don't know why they have kept rest of the videos hidden and publishing 1 video every single day. They could just put the entire playlist at once but anyway I'm loving it because the content quality is too good
@Shivasai121234 I'm not looking for a job in inference at all I beleive it's for experienced people that's why I'm focusing on ai engineering and inference so that if I ever get an opportunity, I won't miss
Day 4 of Inference Engineering
I finished reading the 1st chapter from Zero to SGLang. This was actually a good intro to LLM. I'm very excited for the next chapter that is the intro to Inference
See you tomorrow