12 WEEKS Self-paced · Certificate with TechWave Global (USA)

Become an AI Engineer.

Learn to build real AI products with Claude, GPT, Gemini, DeepSeek and open models: LLM APIs, prompt and context engineering, RAG, agents and MCP, evals, fine-tuning, voice AI and production deployment. Finish with a portfolio, a verifiable certificate, and a guaranteed interview with us when you score 80% or more, plus placement assistance.

₹6,000/ monthAUTOPAY · CANCEL ANYTIME
Try Module 1 free

GST inclusive. Renews every month by UPI autopay or card through Razorpay. Cancel anytime from your dashboard.

Meet Vikram: the whole program in 3½ minutes
11modules
50narrated lessons
11hands-on projects
~107 hof study & labs
12weeks, self-paced
1verifiable certificate

In collaboration with TechWave Global LLC, USA. Your certificate is issued by Indrashikha together with TechWave Global LLC, a US company registered in Washington State, and carries both organisations' signatures and a verifiable QR code.

How it works

From enrolment to interview

Study at your own pace, about 8 to 10 hours a week. Every lesson has an audio explainer, a written lesson with runnable code, and a short quiz.

  1. 1

    Subscribe

    ₹6,000 a month on autopay. Instant access on any device; cancel anytime.

  2. 2

    Learn

    50 lessons across 11 modules, each with a 2-minute narrated explainer, a written lesson, code and a 3-question quiz.

  3. 3

    Build

    One hands-on project per module, ending with a capstone you can show employers on GitHub.

  4. 4

    Exit exam

    Opens after 80% of lessons. 30 scenario questions in 45 minutes, up to 3 attempts.

  5. 5

    Certificate

    Score 60% or more for a verifiable certificate from Indrashikha and TechWave Global LLC (USA).

  6. 6

    Interview

    Score 80% or more for a guaranteed interview with Indrashikha, plus placement assistance.

Curriculum

11 modules, 12 weeks

Designed with the best current models and the way AI products are actually built in 2026. Module 1 is free to preview.

Open Module 1 free
1

How LLMs Really Work

Explain how an LLM turns text into tokens, vectors and predictions, choose sampling and context settings deliberately, and design around hallucination with grounded, verifiable answers.
Week 1 · 5 lessons
FREE PREVIEW
  1. Tokens, not words25 min
  2. Embeddings and attention: how a model reads context30 min
  3. How models are trained: pretraining, instruction tuning, RLHF and RLAIF25 min
  4. Sampling and context windows25 min
  5. Hallucination and what models can and cannot do30 min
Project: LLM X-ray: a measured profile of a model for an Indian multilingual use case · about 4 h
2

Python for AI Engineering

Set up secure, reproducible Python projects and build a tested, async FastAPI backend that calls LLMs through typed, validated interfaces.
Weeks 1-2 · 5 lessons
  1. Project setup: uv, virtual environments and secrets25 min
  2. HTTP and async: httpx and asyncio for LLM calls30 min
  3. Data models with Pydantic and JSON30 min
  4. FastAPI basics for AI backends35 min
  5. Testing AI code with pytest30 min
Project: Support Triage Service: a tested FastAPI backend for an LLM feature · about 6 h
3

Working with LLM APIs

Call Claude, GPT, Gemini and DeepSeek from production-quality Python code with multi-turn history, streaming and validated structured outputs, routed by task and made resilient with retries, fallbacks and caching.
Weeks 2-3 · 5 lessons
  1. Your first calls with the Anthropic and OpenAI SDKs25 min
  2. Messages, roles, system prompts and multi-turn conversations30 min
  3. Streaming responses25 min
  4. Structured outputs: JSON Schema and strict tool calls35 min
  5. Multi-provider routing, rate limits, retries and caching35 min
Project: Multi-provider Smart Inbox API · about 8 h
4

Prompt & Context Engineering

Write, test and version production prompts, choose the right reasoning approach, and engineer a cache-friendly context window for real applications.
Weeks 3-4 · 4 lessons
  1. Anatomy of a strong prompt, with examples30 min
  2. Reasoning: chain-of-thought, thinking modes and decomposition30 min
  3. Context engineering, long documents and prompt caching35 min
  4. Prompt versioning, testing and iteration30 min
Project: Production prompt kit for a Hinglish support assistant · about 5 h
5

Embeddings, Vector Search & RAG

Build, evaluate and improve a production-style RAG system with sensible chunking, a real vector database, verifiable citations and advanced retrieval techniques.
Weeks 4-5 · 5 lessons
  1. Embeddings and similarity25 min
  2. Chunking strategies30 min
  3. Vector databases: Chroma and pgvector35 min
  4. The RAG pipeline end to end, with citations35 min
  5. Advanced RAG and when RAG fails35 min
Project: Cited Q&A assistant over real Indian public documents · about 6 h
6

Tool Use, MCP & AI Agents

Build tool-using agents and MCP servers, choose the right agent architecture, and ship them with permissions, budgets and human approval built in.
Weeks 6-7 · 5 lessons
  1. Function and tool calling30 min
  2. The agent loop35 min
  3. Model Context Protocol: servers and clients35 min
  4. Planning, memory and multi-agent patterns35 min
  5. Agent safety: permissions, human-in-the-loop and budgets35 min
Project: Safe operations agent with a custom MCP server · about 7 h
7

Evals, Testing & Observability

Measure an LLM system with a versioned eval dataset, code and LLM-judge graders, production tracing, guardrails and a CI regression gate.
Weeks 7-8 · 4 lessons
  1. Why evals: from vibes to numbers25 min
  2. Building eval datasets that reflect reality30 min
  3. Code-graded evals and LLM-as-judge35 min
  4. Tracing, guardrails and regression tests in CI35 min
Project: Eval suite and CI gate for your support bot · about 5 h
8

Open Models, Fine-tuning & Local Inference

Choose, run and quantise open-weight models locally, decide rationally between prompting, RAG and fine-tuning, fine-tune with QLoRA and serve the result with vLLM at a known cost.
Weeks 8-9 · 4 lessons
  1. The open-weight model landscape25 min
  2. Running models locally: Ollama, llama.cpp and quantisation30 min
  3. Fine-tune, prompt or RAG? Making the call30 min
  4. LoRA and QLoRA walkthrough, serving with vLLM, GPU costs35 min
Project: Local model versus API, then a QLoRA fine-tune · about 6 h
9

Multimodal & Voice AI

Build features that read images and documents, transcribe and speak Indian languages, hold low-latency voice conversations and generate on-brand images responsibly.
Weeks 9-10 · 4 lessons
  1. Vision: images, documents and OCR-style extraction30 min
  2. Speech-to-text and text-to-speech for Indian languages30 min
  3. Realtime voice agents35 min
  4. Image generation: basics, use-cases and responsibility25 min
Project: Voice-enabled invoice helpdesk for a small business · about 6 h
10

Shipping AI to Production

Design, deploy and operate a secure, cost-controlled and legally aware LLM application, from queue-backed backend to container in the cloud.
Weeks 10-11 · 5 lessons
  1. Architecture of an AI product30 min
  2. Deployment: Docker, serverless and the cloud35 min
  3. Security: prompt injection, data leakage and secrets35 min
  4. Cost control and latency engineering30 min
  5. Responsible AI and Indian law basics (DPDP Act 2023)30 min
Project: Production-harden and deploy an AI service · about 8 h
11

Capstone & Career

Scope, build, evaluate and present a portfolio-grade AI capstone, and prepare for AI Engineer interviews and a focused job search.
Weeks 11-12 · 4 lessons
  1. Choosing and scoping your capstone30 min
  2. Building and documenting: README, demo video, eval report30 min
  3. AI Engineer interview prep35 min
  4. Portfolio, GitHub and the job search in India and abroad30 min
Project: Capstone: ship a portfolio-grade AI application · about 20 h
What you will work with

The models and tools teams use

ClaudeOpenAI GPTGeminiDeepSeek LlamaMistralHugging FacePython FastAPIDockerAzureCloudflare
Sample Indrashikha AI Engineer Program certificate with signatures, gold seal and QR verification
Sample certificate · learner name and number masked
Our promise

A certificate that means something

  • Indian + US collaboration. Certificates are issued by Indrashikha in collaboration with TechWave Global LLC, a US company registered in Washington State, and signed by both organisations.
  • Verifiable certificate. Pass the exit exam with 60% or more and receive a certificate with a unique ID and QR code that anyone, including employers, can verify on indrashikha.com.
  • Guaranteed interview. Score 80% or more in the exit exam and you are guaranteed an interview with Indrashikha.
  • Placement assistance. We help with your resume and portfolio and refer strong learners to companies in our network.
  • A real portfolio. 11 projects and a capstone on your GitHub, which is what interviewers ask about.
  • Interview preparation built in. Module 11 covers LLM system design, coding rounds and common AI Engineer questions.

The interview is guaranteed at 80% or more; a job offer depends on how you do in it. Placement assistance means support and referrals, not a guaranteed job.

Fees

One plan. Everything included.

MONTHLY · AUTOPAY · CANCEL ANYTIME
₹6,000/ month
GST inclusive · renews monthly until you cancel
Secure autopay by Razorpay · UPI AutoPay · Cards
  • All 11 modules and 50 lessons, with audio explainers
  • Runnable code in every lesson, and 11 hands-on projects with a capstone
  • Quizzes after every lesson to check your understanding
  • Exit exam with 3 attempts
  • Verifiable certificate at 60% or more, with TechWave Global LLC (USA)
  • Guaranteed interview at 80% or more, plus placement assistance
  • Full access, including course updates, while your subscription is active
  • Email support at contact@indrashikha.com
Questions

Frequently asked

Who is this program for?

Graduates, final-year students and working developers who can already write basic code in any language. Module 2 brings your Python up to the level the rest of the course needs.

How long does it take?

It is planned as 12 weeks at about 8 to 10 hours a week. It is self-paced, so you can go faster or slower.

Is it live or recorded?

It is self-paced. Every lesson has a narrated audio explainer, a written lesson with code, and a quiz, so you can study at any time on your phone or laptop.

Do I need paid AI API keys?

No. You can use the free tiers of the providers, or run open models locally with Ollama. Lessons show you how to keep costs to a few rupees.

Who issues the certificate?

Indrashikha Educational (OPC) Pvt. Ltd, India, in collaboration with TechWave Global LLC, a US company registered in Washington State (UBI 605 626 585). The certificate carries signatures from both organisations, a gold seal, and a QR code and unique ID that anyone can verify online.

How does the certificate work?

After you finish 80% of the lessons, the exit exam opens: 30 questions in 45 minutes, with up to 3 attempts. Score 60% or more to get your certificate with a unique ID that anyone can verify online.

How does the guaranteed interview work?

If your best exit exam score is 80% or more, an interview form opens in your dashboard. Fill it in and our team schedules an interview with Indrashikha; every learner who qualifies gets one. Through placement assistance we also help with your resume and portfolio and refer strong learners to companies in our network. A job offer depends on your interviews; it is not guaranteed.

How does the monthly subscription work?

You approve a UPI AutoPay or card mandate in Razorpay. ₹6,000 is charged now and then once a month, until you cancel. You can cancel anytime from your course dashboard; access continues until the end of the month you have paid for. Module 1 is free, so you can try before you subscribe.

What happens if a monthly payment fails?

Razorpay retries it and emails you. If it still fails, your access pauses after a short grace period, and your progress is kept. Subscribe again with the same email to continue where you left off.