100 Days of building phases explained!
This is the blueprint of phases that I am going to follow for the challenge
Research Software Engineer / AI Engineer specializing in computational neuroscience, NLP, and healthcare AI.
Phase 1 — Engineering Foundations (Days 1–25)
Goal: Learn how the tools you already use actually work.
Projects:
- ✅ Build a Neural Network
- Build a Git clone
- Build a Shell
- Build a Memory Allocator (optional but valuable)
- Build a Command-Line Argument Parser
You’ll learn:
- Linear algebra in practice
- Backpropagation
- File systems
- Processes
- Unix philosophy
- Clean project structure
⸻
Phase 2 — Systems Programming (Days 26–45)
Goal: Understand what happens underneath Python.
Projects:
- Build an HTTP Server
- Build a Database
- Build a Cache
- Build a Load Balancer
- Build a Key-Value Store
Topics:
- TCP/IP
- HTTP
- Threads
- Concurrency
- Storage engines
- B-Trees
- Hash tables
These are the foundations behind many AI serving systems.
⸻
Phase 3 — AI Infrastructure (Days 46–65)
Now everything starts connecting.
Projects:
- Build a Vector Database
- Build a Search Engine
- Build a RAG Pipeline
- Build a Tokenizer
- Build a Transformer from scratch
Topics:
- Embeddings
- FAISS-like indexing
- Retrieval
- Attention
- Transformers
This phase is especially relevant to your dementia NLP research.
⸻
Phase 4 — Research Engineering (Days 66–85)
Now build tools researchers actually use.
Ideas:
- Experiment Tracker
- Training Framework
- Dataset Versioning Tool
- Annotation Tool
- Medical Image Viewer
- Grad-CAM Visualizer
- Paper Search Engine
This is where your work becomes directly applicable to your lab.
⸻
Phase 5 — Production AI (Days 86–100)
Take everything you’ve built and make it deployable.
Projects:
- FastAPI inference server
- Dockerized deployment
- CI/CD pipeline
- Kubernetes basics
- Monitoring
- Logging
- Model versioning
⸻
Keep one long-term flagship project
Alongside these smaller projects, keep improving one project throughout the 100 days. Given your interests, that could be:
NeuroLab
- Medical image viewer
- NLP pipeline for dementia forums
- Experiment tracking
- Grad-CAM visualization
- Paper search
- Vector search
- AI assistant for researchers
This becomes a portfolio piece that grows as your skills grow.
⸻
Suggested weekly rhythm
Instead of measuring success by finishing projects, measure it by consistent progress.
Monday–Thursday
- Learn a concept.
- Implement a feature.
- Write tests.
- Commit to GitHub.
Friday
- Refactor.
- Improve documentation.
- Add benchmarks.
Saturday
- Write a blog post explaining what you learned.
Sunday
- Review the week.
- Plan the next one.
⸻
How each project should be organized
Every repository should look something like:
project/ │ ├── src/ ├── tests/ ├── docs/ ├── examples/ ├── benchmarks/ ├── assets/ ├── README.md ├── LICENSE ├── pyproject.toml (or equivalent) └── .github/ └── workflows/
This mirrors how many mature open-source projects are structured.
⸻
Build your public profile as you go
For each project:
- A clear README explaining what you built and why.
- A short architecture diagram.
- A blog post summarizing the key ideas.
- Clean commit history.
- Issues or TODOs for future improvements.
Over 100 days, you’ll accumulate not just code, but evidence of how you think and communicate as an engineer.
⸻
A roadmap tailored to your goals
Days Theme Flagship project 1–25 Engineering fundamentals Neural Network 26–45 Systems programming Database + HTTP Server 46–65 AI infrastructure Search Engine + RAG 66–85 Research engineering NeuroLab tools 86–100 Production AI Deploy and integrate everything
If you stick with this plan, by Day 100 you’ll have a portfolio that tells a coherent story: someone who understands AI from the mathematical foundations through to systems engineering and can build research-grade software. That combination aligns well with the kind of computational neuroscience and AI research roles you’re aiming for.