Day 90 — the wrap-up

Day 90.

What I'm taking with me:

- Public learning makes you a sharper engineer. The version of me that explains is more rigorous than the one that consumes.
- Tiny end-to-end projects > theory chapters. 100 lines that run beat 100 pages that don't.
- The default tools matter more than people admit — uv, ruff, AdamW, LightGBM, LangGraph, Qdrant. Pick well once.
- Most production AI is 60% data + 30% engineering + 10% model.
- The community shows up if you do. Thank you.

If this series helped you, the best thank-you is start your own.

Weeks 1-2 — Python: setup, types, OOP, decorators, generators, context, types, errors

Weeks 3-4 — DSA: Big-O, arrays, strings, lists, hash, stacks, recursion, search, sort, trees, graphs, DP

Week 5 — Data: NumPy, broadcasting, pandas, EDA, cleaning

Week 6 — ML: framing, regression, classification, trees, boosting, clustering, SVM

Week 7 — DL: neurons, backprop, PyTorch, CNNs, RNNs, regularisation, training tricks

Week 8 — NLP: tokens, embeddings, attention, transformers, BERT, GPT, fine-tuning

Weeks 9-10 — RAG: chunking, vectors, retrieval, rerankers, query rewriting, multi-hop, agentic, GraphRAG

Week 11 — Agents: ReAct, tools, planning, multi-agent, frameworks, memory

Week 12 — Automation: scraping, browsers, schedulers, notifications, files, APIs, serverless

Week 13 — Career: portfolio, OSS, resume, interviews, learning loop

Full index pinned on my GitHub.

All 450 posts' code snippets are going into one open repo. Searchable. Each post links to its commit. You can clone the whole thing and run any snippet.

Hand-on-keyboard learning beats reading. If you've followed for 90 days, the repo is where you turn 'I know this' into 'I can do this'.

If this series gave you something, three asks:

1. Pick ONE concept and ship a 30-line repo. Tag me.
2. Share the post that helped you with one friend who could use it.
3. Tell me what to cover next — DM open.

The series scales beyond me only because of you. The next 90 days, I'm picking topics from what you ask. Let's go.

Last post.

To everyone who reposted, commented, saved, DMed, or just quietly read — thank you. This sprint was an experiment in public. You made it work.

I'll keep posting. Less daily, more deeply. The 90-day archive stays online. The repo goes public this weekend.

If you started your own learning sprint because of this — please tell me. That's the win.

Saurav.
linkedin.com/in/sauravdnj · github.com/SauravDnj · @saurav_dnj_24