powering

TELERIK ACADEMY SPRINT

Coding with AI agents

Learn to delegate real development tasks without losing control. We'll cover:

👉 Hand AI agents complete tasks, not code snippets

👉 Stop re-explaining your codebase in every session

👉 Verify the output without reviewing every line

👉 Run agents longer without losing control

Victor Valtchev

🎯 AI Engineer and Co-founder, Keen Agents

Enroll now

The Sprint experience

1. Designed to fit your busy schedule.

2. Learn, apply in class, and discuss with trainers and peers.

3. Earn a Telerik Academy certificate to share with your network.

Trainers & Guest speakers

Meet your trainer

Victor Valtchev Trainer

AI Engineer and Co-founder,
Keen Agents

Victor Valtchev is an AI engineer, trainer, and co-founder of Keen Agents, a Sofia-based AI agent platform. As the company’s initial CTO, he has hands-on experience using coding agents in production, coordinating multiple agents, and verifying their work.

Previously a Java and JavaScript trainer at Telerik Academy, Victor now helps development teams adopt agentic coding practices. His approach is tool-agnostic and applies across Claude Code, Codex, Cursor, Hermes Agent, and other coding agents.

Achievements:

Delivered an AI session to 170 participants at Unimasters
Designed and delivered a full-day agentic AI training for 28 developers
Co-founded Keen Agents and led the development of its AI agent platform
Trained professional Java and JavaScript cohorts at Telerik Academy

Contributed to the success of:

Who it's perfect for

Developers new to agentic coding

You already use AI chat or autocomplete but want to delegate complete tasks without rewriting the result from scratch.

Tech leads

You need to decide where human review belongs when agents write code and how to introduce stronger controls across a team.

Experienced developers

You want a repeatable method for increasing task size, setting checkpoints, and reviewing agent work efficiently.

Developers working in any stack

The Sprint is language- and tool-agnostic. You need programming experience, but no previous experience with coding agents is required.

Business value for companies

Investing in this course equips employees with the skills to:

Delegate larger development tasks while keeping human ownership of the result.
Reduce repeated setup and prompting through clearer, reusable repository context.
Scale AI-assisted development more safely with review gates, tests, linting, type checks, and controlled permissions.
Catch incorrect implementations earlier through explicit acceptance criteria and stop conditions.
Build a shared approach to agentic coding that transfers across tools and teams.

What you'll learn

Move from prompting an AI for code snippets to directing an agent through complete, verifiable development tasks.

Prepare agents to work in your repository

Understand how Claude Code, Codex, Cursor, and other coding agents work.
Structure project documentation so agents find the right context and rules.
Build reusable context, checklists, and memory for future tasks.

Define the task. Verify the result.

Set clear goals, guardrails, acceptance criteria, and stop conditions.
Place human review gates based on risk and blast radius.
Verify results using tests, type checks, linting, diffs, and exploratory review.

Scale delegation safely

Increase task size and autonomy only when previous results support it.
Control file access, commands, writes, permissions, and blast radius.
Coordinate longer tasks and multiple agents without losing human oversight.

Curriculum

Detailed curriculum

Across four live sessions, you’ll set up a coding agent, prepare it to work in your repository, add review and verification controls, and plan a supervised run for a real development task.

01Session 1
02Session 2
03Session 3
04Session 4

Give the agent the right context

Get started with agentic coding tools

Compare Claude Code, Codex, Cursor, and other agentic coding tools.
Understand the model, tool calls, and loop behind each interface.
Choose the model, reasoning effort, and task scope for the work at hand.

Write documentation an agent can use

Treat context as a limited budget and recognize when a session has lost the plot.
Organize project knowledge so each fact has one authoritative place.
Help agents find the right files, rules, commands, and failing tests.

Build reusable project context

Set up context files, checklists, memory, and project rules.
Use the provided starter kit and onboarding message.
Decide what to update after every session and task.

After the session: Complete Home Build A on your own small project, allowing the agent to work with minimal intervention. Record time to first working result, interventions, failed checks, and defects found.

You’ll leave with: The starter kit and a method for building reusable project context.

Add human review gates

The session opens with a debrief of Home Build A and the group’s real results.

Place gates based on risk

Identify tasks that are customer-facing, difficult to reverse, or involve secrets.
Decide when to approve the plan, review the diff, or stop the run.
Use checkpoints as controls rather than guarantees.

Use machine checks as the critic

Apply tests, type checks, linting, diffs, and revert mechanisms.
Understand the limits of agent self-critique and when to use an external verifier.
Combine scripted checks with an exploratory review of the working result.

Increase task size safely

Start with a small batch, review it, explain what worked, and then expand.
Write checkpoints before the agent begins.
Review the actual diff and keep a clear path to revert.

After the session: Complete Home Build B on the same project, this time with your own checkpoints and verification plan. Compare the results with Build A.

You’ll leave with: A checkpoint template and verification checklist.

Define the end goal

The session opens with a debrief of Home Build B and the before-and-after results.

Write goals that prevent the wrong solution

See what a vague instruction such as “make the tests pass” can legitimately permit.
Structure every instruction around a goal, guardrails, and a done-check.
Learn from examples based on real agent runs.

Structure the case before the build

Develop requirements through iterations rather than relying on one long prompt.
Create lightweight business and functional requirements an agent can use.
Treat acceptance criteria as the contract for the result.

Move from one task to a project

Define goals that remain useful beyond a single session.
See how several agents can divide one goal while one reviewer keeps ownership of the decisions.
Understand the cost of skipping important check-ins.

You’ll leave with: A goal template covering the goal, guardrails, and done-check, plus a lightweight requirements outline.

Run agents with supervised autonomy

Choose the right level of autonomy

Work across four levels: approve every step, approve the plan, run unattended between gates, or let the agent wake itself and continue on its own schedule.
Understand what each level requires from you.
Move up only when the checks and previous results justify it.

Run an agent safely

Apply six controls: read-only access, an explicit allowlist, human approval for writes, sandboxing, limited blast radius, and observable results.
Match permissions and controls to the risk of the task.
Understand prompt injection and what review gates can and cannot catch.

Plan your next real run

Review examples of long runs, including where and why they failed.
Rely on controls and recovery mechanisms rather than trust.
Choose one real next task and define its autonomy level and gates before the Sprint ends.

You’ll leave with: A safe-run checklist and a concrete plan for your next agentic development task.

Tech requirements

To participate in the live exercises, you’ll need:

A Windows, macOS, or Linux computer with software installation access.
A stable internet connection, browser, Git, terminal, and code editor.
A coding agent that can edit local files and run commands, such as Claude Code, Codex, Cursor, Antigravity CLI, or Hermes Agent.
An account you can use at home for a personal project.

Browser-based ChatGPT alone is not enough. No API key, local GPU, or locally hosted model is required.

Free plans may work, but usage limits can interrupt longer tasks. A paid plan is recommended, not required. Employer accounts are acceptable only if personal projects are permitted. No company code or data will be used.

What sets us apart

Learn from industry experts with proven career results.

Gain practical insights not found on YouTube, Udemy or Coursera.

Uncover tribal knowledge that is rarely documented and shared.

Format & Schedule

How you’ll learn

Two-week training

Four 2-hour sessions (6-8pm) distributed across two consecutive weeks.

Sprints are designed to fit the schedules of full-time
professionals.

What you’ll get

4 goal-oriented lessons with summaries
2 home builds on your own small project.
Group debriefs using participants’ real results.
A starter kit for reusable repository context.
Templates for checkpoints, verification, goals, requirements, and safe runs.
Dedicated Q&A time with the trainer.
Session recordings and access to the learning materials.
A Telerik Academy certificate.

The more you learn, the more you earn

Enjoy discounts on future purchases

Earn a 5% discount on your next Sprint purchase. Your discount will be automatically applied when you sign up for your selected Sprint.

Refer friends and earn even more

For each referral, you and the new user will both receive a 5% discount off the price of a Sprint, which you can use towards your next Sprint purchase. Additionally, the new user will receive an extra 5% discount on their first purchase.

Enroll now

Inspiring passion for learning.
Changing lives and businesses.

Dimitar Markov

DevOps Engineer

SAP

(Class of 2020)

Telerik Academy is a great starting point for anyone who wants to jumpstart a career in the IT industry. It provided me with the underlying knowledge and practical experience needed to join the sector.

Stoyan Damov

Co-Founder

Tick42

The long wait is finally over. Telerik Academy created a brand new upskill program for the world's most popular and loved UI library - React. We trust this new program will help developers gain enough practical knowledge of React, and will significantly cut the time and effort we spend on cross-training internally at Tick42.

Vasil Dininski

Web Developer

Amazon Development Center

(Class of 2013)

Telerik Academy is creating a society of life-long learners. It teaches a mindset that is not only of great help when solving complex problems, but also serves as a wonderful motivator to constantly be hunting for and adopting new skills and information.

Alexander Popov

Site Engineering Lead

Uber

Some of my best junior software engineers at Uber come from Telerik Academy. It is always a pleasure to work with their alumni!

Ivan Rusev

Head of Software Development

Experian

We partner with Telerik Academy because we know their graduates have gone through an extremely demanding program - with high training standards and practice - and have proven themselves to be true technology ninjas.

Kiril Peyanski

Principal Software Engineer

Progress

(Class of 2017)

Telerik Academy changed my life. At the age of 20, I was already independent and working my dream job! Without the knowledge and experience I gained at Telerik Academy, I wouldn’t have had such a flying start.

Still have questions?

Do I need previous experience with coding agents?

No. You should be comfortable working as a developer, using Git, a terminal, and an IDE, but you can be new to agentic coding.

Which coding tool should I use?

Claude Code or Codex is preferred, but Cursor, Antigravity CLI, Hermes Agent, or another tool that can read and write local files and run commands will work. The Sprint teaches a transferable method rather than one product.

Is a paid AI plan required?

No, but it is recommended for the home builds. Free plans may have limits that interrupt a longer task. Paid plans also have usage limits.

Can I use code from work?

No. You’ll use your own small project outside work so no employer code, data, or confidential information is involved.

How much work is required outside the live sessions?

You’ll complete two required home builds after Sessions 1 and 2. The time needed depends on the project and task you choose, so no fixed duration is promised.

How do I enroll for a Sprint?

Simply click the enroll button and follow the on-page instructions. You’ll need to sign up for a Telerik Academy Learn account to access Sprint session links and course materials. Then, just pay for your seat to complete your enrollment. That's it!

How can I pay?

You can pay for your seat using your Visa, Mastercard, American Express or Diners Club card directly on our website. If you prefer to make a bank wire, please contact us for assistance.

Do you offer group discounts?

We do! For more information, please get in touch with us.

Can I get an invoice?

To receive a company invoice, check the "I want an invoice" box during sign-up and provide your company's name, address, and VAT number. We'll email the invoice to you.

How long does it take to complete a Sprint?

Sprints are 2-week programs with 4 sessions held on 2 weekdays each week, scheduled for after-work hours.

What if I miss a live session?

No problem! We record every live session with our trainers and speakers and make the recordings available for you to replay at your convenience.

Will I receive a certificate upon completion of a Sprint?

Every trainee who completes their Sprint earns a Telerik Academy certificate of completion. To qualify, you must attend or watch on demand all course content. You can print the certificate or add it to your LinkedIn profile. Feel free to tag @Telerik Academy when sharing it.

How is a Sprint different from an Upskill?

Sprints are 2-week, hands-on programs where industry leaders share practical, ready-to-use knowledge that's immediately applicable at work. Upskills, on the other hand, provide a comprehensive, 3-month journey to bridging skill gaps in a specific domain, helping you advance to the next career level.

How can I get in touch?

Just send us an email and we'll get back to you soon.

Ready to get started?

Enroll now

Coding with AI agents

Go beyond AI snippets. Delegate complete development tasks.

The Sprint experience

Meet your trainer

Who it's perfect for

Business value for companies

What you'll learn

Detailed curriculum

Give the agent the right context

Add human review gates

Define the end goal

Run agents with supervised autonomy

Tech requirements

What sets us apart

How you’ll learn

The more you learn, the more you earn

Inspiring passion for learning. Changing lives and businesses.

Still have questions?

Do I need previous experience with coding agents?

Which coding tool should I use?

Is a paid AI plan required?

Can I use code from work?

How much work is required outside the live sessions?

How do I enroll for a Sprint?

How can I pay?

Do you offer group discounts?

Can I get an invoice?

How long does it take to complete a Sprint?

What if I miss a live session?

Will I receive a certificate upon completion of a Sprint?

How is a Sprint different from an Upskill?

How can I get in touch?

Ready to get started?

Thank you for subscribing!

Get the latest from Telerik Academy

School School

Alpha Alpha

Upskill Upskill

Sprint Sprint

For Business For Business

About About

Inspiring passion for learning.
Changing lives and businesses.