AI ENGINEERING PORTFOLIO ROADMAP

Don't just
learn AI.
Build your proof.

From “I followed the tutorial”
to “Let me show you how it works.”

Start with Python. Build toward three deployed AI systems with evidence, guardrails and human approval. Learn the decisions behind the code—not only the code.

Explore the builds

₹18,999 first installment of 4

₹75,996 total course fee. Remaining installments invoiced separately.

Published cohort: 24 August 2026 – 20 November 2026. Joining now? Confirm catch-up and payment timing before paying.

your-ai-portfolio /
NOT JUST ANOTHER NOTEBOOK.Your next
engineering story.
knowledge_copilot/Answers with evidence01
control_mapper/Structure with guardrails02
governed_agent/Actions with accountability03
ScopeBuildEvaluateShip
Bring your code.
Be ready to explain your decisions.
COURSE BUILD PREVIEWIllustrative—not student results
Your repository.Your reasoning.
03project builds in the curriculum
26live working sessions in the full program
13 weeksPython foundations to portfolio launch
Mon + Fri9–10 PM Indian Standard Time (IST)

THREE SYSTEMS. THREE ENGINEERING CONVERSATIONS.

A portfolio with
something to say.

Explore the planned builds. The interesting part is not the interface—it is the evidence and judgment underneath it.

BUILD DOSSIER / 01Planned learning artifacts

Compliance Knowledge Copilot

Can your answer show its evidence?

  1. 01Approved sources
  2. 02Hybrid retrieval
  3. 03Evidence gate
  4. 04Cited response

When the sources cannot support an answer, the system should say so—not improvise.

Weeks 5-8 (Sessions 10-16) · Your implementation, documentation and demo

The first build uses retrieval-augmented generation (RAG). The mapper uses JavaScript Object Notation (JSON) schemas. The agent introduces human-in-the-loop (HITL) approval.

Build the system. Learn to explain why.

THE VALUE IS IN THE WORK YOU LEARN TO DO.

Information is everywhere.
Build capability.

The fee is for a structured, live build process: a sequenced curriculum, scoped independent work, evaluation and a path to demonstration. Its value depends on doing the work—not simply attending.

01

Code you can navigate.

Your own repositories, structured components and a system you have worked on—not just a copied final notebook.

02

Quality you can question.

Evaluation sets, schema checks and failure analysis that help you explain where the system works and where it does not.

03

Decisions you can defend.

Architecture diagrams and documented trade-offs that turn “I used this tool” into a reasoned engineering conversation.

04

Work you can demonstrate.

Deployed demonstrations, project READMEs and demo videos that make your implementation inspectable.

Instead of only saying
“I know how AI agents work.”

Work toward being able to say
“Here is my agent, its approval boundary,
and the trace that explains its action.”

A SEQUENCE. NOT A TOOL COLLECTION.

Start with the basics.
Keep building.

Six foundation sessions, then a connected engineering journey. Every stage has a reason—and work that carries into the next.

Mon & Fri, 9:00 - 10:00 PM IST4-6 hours between sessions of independent work between sessions.Dates below are the published 24 August 2026 – 20 November 2026 cohort calendar.
01Weeks 1-3 · 6 sessionsPython, NumPy and Pandas Foundations

Start from the language itself. Six working sessions that take you from Python syntax to vectorised NumPy and real Pandas analysis, so every later session is about AI engineering and never about fighting the basics.

  • Write clean Python with functions, types and error handling
  • Structure a project with virtual environments, Git and the command line
  • Use NumPy arrays and vectorised operations instead of loops
  • Load, clean, group and analyse real data with Pandas
  1. 01
    Python Foundations

    Environment setup, variables, types, control flow and the command line

    Mon, 24 Aug
  2. 02
    Functions & Data Structures

    Lists, dicts, comprehensions, functions, modules and error handling

    Fri, 28 Aug
  3. 03
    Files, APIs & Project Structure

    JSON and CSV, calling APIs, virtual environments, Git and repository layout

    Mon, 31 Aug
  4. 04
    NumPy Arrays

    Arrays, dtypes, indexing, broadcasting and why vectorisation beats loops

    Fri, 4 Sept
  5. 05
    Pandas DataFrames

    Loading data, selection, filtering, joins and reshaping

    Mon, 7 Sept
  6. 06
    Data Cleaning & Analysis

    Missing values, types, grouping, aggregation and your first kept analysis

    Fri, 11 Sept
02Weeks 4-5 · 3 sessionsFoundations

Set the portfolio strategy and the engineering discipline that every later session depends on: scoping, prompting, structured outputs and measurable evaluation.

  • Scope a portfolio that signals real engineering ability
  • Design prompts that fail predictably instead of randomly
  • Score model output against a golden set instead of guessing
  1. 07
    Portfolio Blueprint

    Scope, success criteria, architecture, repositories and deployment plan

    Mon, 14 Sept
  2. 08
    Prompt Engineering

    Few-shot design, roles, constraints and failure modes

    Fri, 18 Sept
  3. 09
    Structured Outputs & Evals

    JSON schemas, validation, golden set and baseline scoring

    Mon, 21 Sept
03Weeks 5-8 · 7 sessionsProject 1 - Compliance Knowledge Copilot

Build and deploy a retrieval system that answers compliance questions with control-level citations, and prove its quality with a real evaluation harness.

  • Build a licensed, traceable document corpus
  • Run hybrid retrieval with reranking and debug it properly
  • Answer with citations and refuse when evidence is insufficient
  • Report retrieval metrics and faithfulness, not vibes
  1. 10
    Corpus & Licensing

    Approved frameworks, source manifest and metadata schema

    Fri, 25 Sept
  2. 11
    Document Ingestion

    Parse, clean and chunk while preserving control IDs and source links

    Mon, 28 Sept
  3. 12
    Indexing Pipeline

    Embeddings, vector database, keyword index and metadata filters

    Fri, 2 Oct
  4. 13
    Retrieval Quality

    Hybrid search, reranking, top-k experiments and retrieval debugging

    Mon, 5 Oct
  5. 14
    Grounded Answers

    Control-level citations, source links and insufficient-evidence handling

    Fri, 9 Oct
  6. 15
    RAG Evaluation

    30-50 questions, retrieval metrics, faithfulness and failure analysis

    Mon, 12 Oct
  7. 16
    Copilot Deployment

    UI, API, cloud deployment, README, architecture diagram and demo

    Fri, 16 Oct
04Weeks 9-10 · 4 sessionsProject 2 - Control Mapper

Turn unstructured risk descriptions into reliable structured output, then harden it with schema checks, retries, confidence scoring and adversarial tests.

  • Model a control taxonomy with evidence and ownership
  • Decompose extraction tasks for reliable structured output
  • Add guardrails, retries and explicit confidence
  • Ship with a test suite and deployed demonstration
  1. 17
    Control Taxonomy

    Define risks, applicable controls, evidence, owners and next actions

    Mon, 19 Oct
  2. 18
    Mapper Prompting

    Task decomposition, examples and reliable structured extraction

    Fri, 23 Oct
  3. 19
    Mapper Guardrails

    Schema checks, retries, confidence, limitations and adversarial tests

    Mon, 26 Oct
  4. 20
    Control Mapper Ship

    Test suite, UI/API, documentation and deployed demonstration

    Fri, 30 Oct
05Weeks 11-13 · 5 sessionsProject 3 - Governed Agent

Build an agent that is deterministic, auditable and safe: a state machine with tools, checkpoints, an audit log and a human approval gate before high-severity actions.

  • Design a deterministic agent workflow with checkpoints and audit logs
  • Automate vendor risk review with explained rationale
  • Triage audit evidence and flag gaps against approved sources
  • Gate high-severity findings behind human approval
  • Evaluate agent runs with traces and structured reports
  1. 21
    Agent Workflow Design

    Deterministic state machine, tools, checkpoints and audit log

    Mon, 2 Nov
  2. 22
    Vendor Risk Review

    Parse questionnaires, map controls, score risk and explain rationale

    Fri, 6 Nov
  3. 23
    Audit Evidence Triage

    Search approved sources, build checklist, flag gaps and risk indicators

    Mon, 9 Nov
  4. 24
    Human Approval Gate

    Review high-severity findings before approve, reject or edit actions

    Fri, 13 Nov
  5. 25
    Agent Evaluation

    Three test runs, structured reports, trace review and follow-up drafts

    Mon, 16 Nov
06Week 13 · 1 sessionPortfolio Launch

Final quality pass across all three projects, then the storytelling that converts a repository into interview offers.

  • Ship polished READMEs and architecture diagrams
  • Record demo videos for each project
  • Tell the engineering story that lands interviews
  1. 26
    Portfolio Launch

    Final QA, README polish, demo videos and interview-ready storytelling

    Fri, 20 Nov
AFTER SESSION 6Python, NumPy and Pandas ready for AI engineering
AFTER SESSION 16Deployed Compliance Knowledge Copilot
AFTER SESSION 20Deployed Control Mapper
AFTER SESSION 25Deployed Governed Agent
AFTER SESSION 263 deployed, evaluated and interview-ready projects

THE HOUR IS STRUCTURED. THE LEARNING IS ACTIVE.

Less watching.
More working.

A live session is the start of the work, not the end. Build a component together, evaluate its behavior, then keep developing it in your own repository.

4-6 hours between sessions of independent work between sessions. Budget for this—not just the live hour.
  1. 0-5 min
    Review the last build

    Review independent work and the one thing that broke most often

  2. 5-15 min
    Understand the decision

    The engineering decision of the day and why it matters in production

  3. 15-50 min
    Build the component

    Build the component together, committed to your own repository

  4. 50-55 min
    Inspect the result

    Evaluate output quality, failure modes and what to harden next

  5. 55-60 min
    Know your next step

    Scoped 4-6 hour task that advances your portfolio project

A SERIOUS INVESTMENT DESERVES A CLEAR FIT.

Bring curiosity.
Make room for practice.

You do not need prior Python experience for the full program. You do need basic programming logic and time to work between sessions.

THIS CAN FIT YOU

You want to move from learning to making.

  • Learning Python and ready for a structured foundation.
  • A software engineer moving into AI systems.
  • A data scientist moving beyond notebooks.
  • A governance, risk and compliance (GRC) professional who wants to build.
PAUSE IF YOU EXPECT

A result without the independent work.

  • Done-for-you projects or passive video consumption.
  • A guaranteed job, salary or employer endorsement.
  • Every planned deployment to be automatic or effortless.
  • All earlier live sessions still to be ahead of you when joining late.

CLEAR FEES. A CONCRETE LEARNING PLAN.

Invest in work
you can stand behind.

Compare the fee with the learning process and the time you can commit—not with a promised salary. The goal is a portfolio you understand deeply enough to demonstrate, question and improve.

  • 26-session curriculum, starting with Python, NumPy and Pandas
  • Three connected builds, from retrieval to governed agents
  • Evaluation, deployment, documentation and demo preparation
  • Scoped independent work between live sessions

Ask about any model-provider or cloud usage costs before enrolling. The course fee is not a promise of unlimited infrastructure.

AI ENGINEERING PORTFOLIO ROADMAP

Begin with the first installment.

₹18,999

4 equal installments · ₹75,996 total

Secure checkout · Sign-in required

  1. 01
    At enrollmentOn or before 24 Aug 2026
    ₹18,999
  2. 02
    Week 414 Sep 2026
    ₹18,999
  3. 03
    Week 812 Oct 2026
    ₹18,999
  4. 04
    Week 112 Nov 2026
    ₹18,999

You pay the first installment to confirm your seat. The remaining three are invoiced by email on the dates above - there is no automatic recurring charge on your card.

Joining after the start? Confirm the catch-up plan and installment timing before paying.

DECIDE WITH YOUR EYES OPEN.

Good questions.
Straight answers.

Not sure about your starting point? Talk to us before enrolling

Is the Agentic AI Portfolio course open for enrollment?

Yes. Enrollment is open and sessions run live on Monday and Friday evenings from 9:00 to 10:00 PM IST, starting 24 August 2026 and finishing 20 November 2026. You can secure your seat with the first installment rather than paying the full fee upfront.

Does the course start from Python basics?

Yes. Weeks 1 to 3 are six live sessions on Python fundamentals, functions and data structures, files and APIs, NumPy arrays and broadcasting, Pandas DataFrames, and data cleaning and analysis. The AI engineering build sessions begin in week 4, so you are never fighting the language while trying to ship a system.

How is this different from a normal AI course?

Most courses end with a notebook. This one ends with three deployed systems: a grounded RAG copilot, a guardrailed control mapper and a governed agent with a human approval gate. Each has an evaluation harness, a README and a demo video, because that is what hiring managers actually inspect.

What do I need to know before joining?

Nothing beyond basic programming logic. The first three weeks are six sessions on Python, NumPy and Pandas, plus Git and the command line, so you have the foundation before the AI build sessions start. If you already write Python confidently, those sessions are a fast refresher and your recordings stay available.

Can I pay in installments?

Yes. The course is paid in four equal installments, and you enroll with the first one.

How much work is expected between sessions?

Four to six hours of independent work between sessions. The live sessions are working sessions, so the building continues in your own time and gets reviewed in the next class.

YOUR NEXT PROJECT CAN BE MORE THAN A DEMO.

Make something
you can explain.

Build deliberately. Test honestly. Tell the engineering story.

₹18,999 first installment · ₹75,996 total course fee