Companies don't hire AI users. They hire engineers who build and deploy production AI systems. 17 weeks. 33 real builds. Every graded one reviewed by an engineer, not a bot.
// Built and taught by Silicon Valley practitioners who deployed AI at scale
// Week 1 free · No card · Cancel any time
Designed and taught by Silicon Valley practitioners who deployed AI at scale.
Recorded weekdays · Live weekend builds · Built for developers who already ship code, and anyone prepared to work like one.Two senior technology leaders looked at the curriculum and put their names to it.
"Most AI courses produce people who can talk about AI. CoreSmart produces people who can actually ship. The curriculum covers the same production-grade stack used in real enterprise systems, and every week delivers a usable artifact rather than just a certificate."
"Teams already have AI tools, but delivery velocity remains unchanged. CoreSmart's dual-track model addresses the real gap by helping Business Analysts specify agent behaviour and Developers govern AI systems at scale."
Recorded content during the week, live build sessions on the weekend. You are never grading yourself — every graded project comes back with a score and written feedback.
Recorded weekday content covering the week's concepts. Watch on your own schedule, before the live session.
The weekend live session. Project 1 is built alongside the instructor, line by line, with the trade-offs argued out loud.
Project 2 — same skills, new input. You ship it yourself with mentor support available throughout the week.
The CoreSmart team reviews your build and returns a scorecard with written feedback before the next week starts.
ReleaseBot turns messy release notes into a structured changelog — streaming over SSE. This is the real thing, running.
// Notice what it actually did: "api is way faster now (~40%)" became a PERF entry with p95 numbers, and "killed the old v1 api" was classified BREAK. Semantic classification, not string formatting.
32 weekly projects across 16 weeks, plus your capstone — 33 real builds. Project 1 you build live with the instructor. Project 2 is yours to ship and gets graded. Same skills, new input. Every one lands on your GitHub.
One real production system — running on real data, not a toy dataset or a staged demo. Choose from six ready-made capstone projects with the data included, or build on your own. Presented and defended live to a panel on Demo Day, with detailed feedback, a score and your certificate.
Every credential below is public and linked. Check them before you read the curriculum.
Open any week to see what you'll learn, the stack you'll use, and the two projects that leave your machine at the end of it.
The whole course comes together in one real production system — not a demo build. Choose one of six ready-made capstone projects — each shipping with a cleaned, licensed dataset — or bring your own data and we help you source and scope it. You present and defend it live to a panel, and receive detailed feedback, a score and your certificate.
Every capstone ships with a cleaned, licensed dataset we provide, so you build from day one. Prefer your own idea or domain? Bring it, and we help you source the data and scope it.
Ask 40,000 real court opinions anything: grounded, cited answers over public US case law.
A patient-info assistant with medical-grade guardrails, grounded in official health sources.
An analyst agent that reads SEC filings and writes cited earnings briefs on any company.
A developer support copilot trained on real Stack Exchange Q&A and open-source docs.
An agent that plans complete day-by-day trip itineraries, cited from real travel guides.
A scheduled agent that turns the week's world events into one clean, cited briefing.
Presenting a working system live, and defending your choices to a panel, is the closest thing to the real job.
// This is skills and positioning support, not a job guarantee.
// These are the titles the work maps to. We don't promise the offer — we make sure you can answer for the build.
We're not an ed-tech company that hired some engineers. We're an AI engineering company that decided to teach.
Building production AI systems, we kept meeting the same developer: shipping real code, using AI daily, and unable to say whether the AI part was correct. Not incapable — just never taught how that question gets answered.
Another course on prompting won't close that. The engineering underneath will: evaluation, guardrails, deployment, reliability and cost. So we teach the playbook our own engineers use, and we grade the work rather than handing out a certificate.
Our mission: turn learners into practitioners, and practitioners into builders.
No prior AI background is required — Week 1 starts with onboarding and fast-track foundations, including a Python and API refresher, to get everyone up to speed. Comfort with basic coding will help, but you don't need to know AI or ML going in.
170+ hours across 17 weeks — roughly 10 a week. Weekday content is recorded, so you fit it around work. The build sessions are live on weekends. Sessions are recorded, so missing one doesn't put you behind.
Partly. The weekday content is recorded, so you learn on your own schedule. The build sessions are live on weekends — that's where Project 1 gets built alongside the instructor, line by line. And you're never grading yourself: every graded project is reviewed by the CoreSmart team and returned with a scorecard and written feedback before the next week starts.
Two real projects every week for 16 weeks, plus a capstone — 33 builds in total, all on your GitHub. Project 1 is built live with the instructor; Project 2 is the same skill applied to a new input, which you ship yourself and which gets graded. They start at streaming services and classifiers, move through RAG, evaluation and fine-tuning, then agents, guardrails, deployment, reliability and cost.
Vinay Bamil designed the course and teaches most of it — a PhD in AI and a former Gen AI Coach at Google. Randeep S. Bhatia advises on architecture and scale, with a background across Twitch, Audible and EA and AAAI-published work. Sanjay Lalwani runs production machine learning at Siemens. Every credential is public and linked on this page — check them before you decide.
Week 17 is capstone week. You take one of six ready-made capstone projects — dataset included — or bring your own data, and build one production-grade system. You then present and demo it live to a panel, graduation-style, and defend your architecture choices. The panel scores your build and your presentation, and you receive detailed feedback and your certificate. Afterwards we review your portfolio and help you prepare to explain everything you built. That's skills and positioning support, not a job guarantee.
Don't take our word for it — check the people. Vinay's Google and PhD credentials, Randeep's Twitch and Audible record, Sanjay's role at Siemens: all public, all verifiable in a few minutes from the links above. Ashish Mago and Jignesh Modi have publicly attached their names to the curriculum. And Week 1 costs nothing and is the real first week of the course, so you can judge the teaching directly instead of judging a sales page.
Full refund within 14 days of payment, no questions asked. After 14 days the fee is non-refundable, since by then you've had the live sessions, the graded reviews and the course material. Week 1 is free either way, so you can see the real teaching before any money moves.
₹85,000 one-time, all taxes included — or three monthly instalments of ₹28,333. Week 1 is free and is part of the 17 weeks rather than an extra taster before it, so you are not charged to attend it. The fee applies only if you continue into Week 2.
Week 1 is free and it is part of the 17 weeks — not a taster built for the brochure. Same live build session, same graded project, same scorecard.
Takes under a minute. We will email you the Week 1 joining details.
