Niralworks Academy·Cohort 01 · 40 seats
Build autonomous AI systems, ship a production-grade portfolio, and get hired for the roles companies are desperately trying to fill — in twelve focused weekends.
Phase 01 build · document-intelligence assistant
Week 3 output: answers with receipts, source pane, traces. This is the first repo in the portfolio you'll defend.
Cohort 01
No fake alumni logos. The proof is the work you ship — starting with the assistant above.
Mentors who ship
Practitioners from fintech, ML platforms, and cloud ops. What they built last month is what they teach.
Hiring-manager test
A public GitHub with traces, evals, and a cost model. That's what gets you the call — not a placement letter.
01·Why this exists
Across fintech, commerce, logistics, and SaaS, companies are deploying agents that plan, call tools, and complete real work. Engineers who can build and operate these systems are scarce enough that hiring managers wait months.
Your university syllabus won't cover any of this before you graduate. Mass-recruiter IT jobs won't either. We will — in twelve focused weekends, taught the way serious engineering is always taught: fundamentals first, then frameworks, then production.
No hype. No prompt-of-the-week. A discipline, learned properly.
The agentic engineer's toolkit — what you'll be hired to do
02·Curriculum
Every phase ends with something that works. Every week ends with a code review from a working engineer.
AG-101
Phase 01 · Weeks 1–3
You'll buildA document-intelligence assistant that answers with receipts.
AG-201
Phase 02 · Weeks 4–6
You'll buildAn autonomous research agent that browses, extracts, and reports.
AG-301
Phase 03 · Weeks 7–9
You'll buildA supervised multi-agent pipeline that completes real work end-to-end.
AG-401
Phase 04 · Weeks 10–12
You'll buildYour capstone — a production agent system, deployed and defended.
Live labs + recorded material · Code review every week · Cohort capped at 40
03·Outcomes
We don't do fake guaranteed placements. We give you the GitHub, the production judgment, and the builder network that makes hiring managers reach out to you.
01
Three or more working agent systems on your GitHub — not tutorial clones, but tools with real users, real failure modes, and real fixes. This is what hiring managers open before they open your résumé.
02
A production agent system with evaluations, tracing dashboards, and a cost model. You'll defend it like an engineering review, not present it like a slideshow.
03
You'll think in architecture decisions, failure modes, and cost tradeoffs — the difference between someone who follows agent tutorials and someone who can own an agent system.
04
A cohort of forty, an alumni community, and mentors who stay reachable after the program ends. The peers you'll build with for years — and the people hiring managers ask for referrals.
The portfolio a hiring manager opens
Roles this work is built for
04·Who it's for
Built for
Not for
Saturday
3-hour live lab with the week's concept, then a build brief on the spot.
Sunday
Build review with mentors — your code, critiqued line by line — plus office hours.
Through the week
Self-work and assignments. Budget 10–12 hours total, at the pace of a real job.
05·Plans
Start with fundamentals, or go straight to production-grade agent engineering. Both are live, mentor-led, and capped.
P·01
8–10 weeks · live online
2nd-year students building real fundamentals
You build and explain a working AI application — and start a GitHub portfolio that grows every year.
Apply for FoundationP·02
Flagship · 40 seats
12 weekends · live online · Sat–Sun
3rd/4th-year students going production-grade
You design, build, evaluate, and deploy production-oriented agentic systems — with a defended capstone.
Apply for 1 of 40 seats06·Mentors
Every mentor works on agentic systems in production. They teach what they shipped last month — not what they read about last year.
Lead Mentor · Agent Platforms
Global fintech · payments
Payment-reconciliation agents in production — millions of transactions a day.
Teaches architecture and the unglamorous parts of production: retries, idempotency, and what breaks when money is on the line.
Mentor · Applied LLMs
Enterprise ML platform
Retrieval and evaluation infrastructure used by internal agent products.
Former ML engineer. Leads Phase 01 and makes fundamentals feel like power tools — citations, structured outputs, context that actually holds.
Mentor · Systems & Deployment
Cloud platform engineering
Operated agent fleets on AWS and GCP, including cost and latency SLOs.
Runs the Ship phase: deployment, cost, latency, and things that break at 2 a.m. You will leave knowing how to keep an agent alive.
Mentor · Evaluations & Safety
LLM evaluation research
Reliability test suites for tool-using agents — pass/fail, not vibes.
Obsessed with measurable reliability. Teaches you why 'it seems to work' is not an eval, and how to defend a system in front of an engineering manager.
Guest sessions from engineers and founders run through every cohort
07·Questions
Anything else? Write to hello@niralworks.com
08·Apply
Apply for 1 of 40 · Cohort 01 · closes Sep 30
No essays, no recommendations. Tell us which plan you want — we reply personally within two working days, then a short call if it's a fit.