AI/ML is the latest dimension of the full stack engineer.
As the stack expands, so does Codesmith’s program to meet the AI/ML revolution.
From September our flagship Software Engineering Immersive will expand by 20% to include AI and ML.
A new technologist, for a changing world
Our world faces unprecedented challenges, and software is the medium that will drive solutions across most domains - from healthcare, finance, cleantech, and government - and AI/ML will multiply that impact.
To meet this moment, it calls for a new kind of technologist — the modern software engineer who is the bridge between people, society, and technology — able to wield AI responsibly.
At Codesmith we’ve cultivated almost 4000 technologists over the last decade and we’re now evolving our program to meet the AI/ML revolution in tech.
Our new curriculum builds on from the work of our Data Science & Machine Learning research group where Codesmith residents contributed to tooling for industry-standard ML tools like Tensorflow.js under the leadership of Codesmith co-founder Alex Zai.
Our approach doesn’t just train technical coders; we build engineering leaders. Folks with the capacity to decide when, where, and how to wield AI/ML to solve real-world problems — with empathy for users — like our alum Victor Luo, who’s now an ML Engineer at TikTok.
James Laff, Senior Curriculum Manager at Codesmith, says the program takes residents from “first principles to implementation to original work.”
“Today, AI ML is another dimension of a well rounded, full stack engineer. The stack has expanded. This is not a pivot from software engineering to AI.
"The purpose of software engineering is solving problems with technology. And now there is a whole other class of tools at your disposal.”
The updated program now includes:
Embeddings & Prompting Heuristics Unit
- Residents develop a rich understanding of how state-of-the-art models represent data through weights and embeddings, learn how to interact with LLMs to produce intended results reliably through prompting, and gain experience integrating AI models into applications.
RAG & Fine-Tuning Unit
- Retrieval-Augmented Generation (RAG) combines information retrieval with GenAI models to enhance accuracy and relevance, providing an effective and efficient way of programmatically scaffolding queries with additional context. Fine-tuning is a powerful tool for adapting models to particular domains, provoking a certain tone, and reducing toxicity and bias.
- Residents build and optimize RAG pipelines, fine-tune models, and develop deep insight into production-level considerations around AI integration.
AI/ML Project
- Residents harness AI to solve a real-world problem by building an application from scratch, choosing and integrating an appropriate model, and establishing reliability through evaluation.
MLOps Lecture
- Residents explore crucial challenges to AI deployment and the evolving ecosystem to address them: orchestration (LangChain), indexing (LlamaIndex), vector DBs (Chroma) and observability (LangSmith).
Residents’ learning journey culminates in open-source product (OSP) development. Having progressed from mental models (pre-program) to implementation (in the units and project), residents are equipped to build open-source AI/ML developer tools. (Residents may choose to focus their expertise on another area of the full stack, depending on their goals and interests — but AI/ML tooling will be an option.)
The best way to grow as an engineer is to build tools for other engineers. Not only do you need to know how to use a given technology — you need to understand the tradeoffs involved with using it, the challenges faced by experienced engineers, and how to mitigate those challenges. The OSP process cultivates mature engineering judgment, as residents move beyond straightforward tasks to challenging design decisions.
The first cohorts to receive the new AI/ML content in-program will be:
- FTRI 51 (Oct 7, 2024 – Jan 17, 2025)
- PTRI 17 (Sep 21, 2024 – Jul 5, 2025)
✨ All current and future alums will have full access to Codesmith’s in-program AI/ML content ✨
Pre-Program
Codesmith is committed to providing the resources aspiring software engineers need. Our free workshops and CSX learning platform help anyone build the capacities to succeed in our rigorous admissions process.
Our AI/ML learning journey begins with a first-principles understanding of how state-of-the-art AI/ML components work — neural networks (data representation, weights and activation functions, gradient descent, backpropagation) and LLMs (tokenization, embeddings, self-attention, and transformer architecture) — as well as how businesses incorporate AI/ML into their products.
We’ve launched new public workshops to help our community establish mental models for working with these new technologies and understand how AI fits into the the structure of modern engineering teams:
- AI Hard Parts: LLMs and Embeddings
- AI for Software Engineers
- Deep Learning & Neural Networks (Part 1)
- Deep Learning & Neural Networks (Part 2)
- How To Stand Out for AI Product Teams (Part 3)
With some of the lectures already in motion, we are seeing how participants are responding well to the new content and Codesmith co-founder, and CEO Will Sentance looks forward to the wider Codesmith community now benefiting from the expansion of the program.
Having recently given a talk on neural networks, he explains how delving into these tools and how they function lifts the lid on what is happening as we use them, a core part of preparing engineers to go on and implement these tools into their future work.
“Under the hood of neural networks means understanding how we’re taking sample data, identifying the patterns within it, and capturing them to make predictions about the broader population,” Will says.
“Laying out that and other mental models illuminates so much of what’s happening when building with and manipulating AI tools.
“Our aim isn’t just around better use of AI tools, but implementing and improving them.”
Stan Louie, an engineer who has followed the ML sessions from James Laff said they “provide a solid bearing at understanding the concepts and technologies.”
“In my previous life prior to CS,” he explains, “I worked on the data side of feeding various ML models, so it is helpful to learn more about them within the SWE context.”
“The possible employment opportunities with ML applications are very exciting and will only expand and deepen. This time around, rather than being adjacent to these technologies, I look forward to building them.”
This focus on building residents’ understanding of these tools for the purpose of implementing them — whether for a tech team's productivity, or to solve complex tech problems — is a facet of the Codesmith AI ML program that sets it apart from other curricula.
Codesmith is committed to helping build an equitable tech landscape, and part of that is ensuring AI and ML tools serve those historically underrepresented in tech, and that the tools don’t become a reflection of one group at the expense of others.
This is why residents will also delve into the ethics behind using these tools, to understand the work needed to ensure AI and ML are devoid of implicit biases, and to give residents a broad perspective that allows them to go on to build products that serve all peoples equally.