Deesha AI · PathwayP-04
AI Application Engineer
Ship AI features with retrieval, tools, agents and evaluation inside a real application.
- Length
- 14–16 weeks
- Courses
- 6
- SkillLabs
- 2
- Mode
- Online / Hybrid
- Fee
- On request
Next intakes · 4 ways in
Start here if you write code and want AI in your daily workflow first.
Reserve a place →Join at the stack if you already have working Python behind you.
Reserve a place →Join at the stack if Spring Boot is home ground.
Reserve a place →
The route
How you get there
5 groups over 14–16 weeks, and 2 of them are places you can join. You leave each one able to do the thing it names — not having sat through it.
01 · Work AI-Native2 weeksJoin here
24 August · Code with Claude · Start here if you write code and want AI in your daily workflow first.
Drive an agentic coding assistant with plans, reviews and guardrails — and learn, from daily use, what models can and cannot do.
02 · Build the Stack4 weeksJoin here
7 September · Build with LLMs — Python · Join at the stack if you already have working Python behind you.
28 September · Spring AI Developer · Join at the stack if Spring Boot is home ground.
Build the complete application stack around a language model in your own language — prompts, structured outputs, tools, memory and deployment.
03 · Ground It in Your Systems3 weeks
Ground answers in your own documents with citations, and connect models to real systems through MCP — built, secured and tested.
04 · Make It Trustworthy2 weeks
Evaluation & AI QualityProve an AI feature is good enough to ship — evaluation datasets, guardrails against injection, and cost and latency you can read.
05 · Give It a Job — Agents3 weeks
Agent Builder LabAgentic AI EngineeringScope, build and bound a tool-using agent — and defend the decision of when an agent beats a workflow.
What you finish with
A deployed AI application with retrieval, tool calling, memory, evaluation gates and observability — built in your chosen stack and defended like a production system.
What the cohort includes
- Cohort-paced structure with milestones, not a self-paced video library
- Mentorship from practitioner trainers who build with these tools daily
- Code review on your own work, every sprint
- Graded capstone with individual feedback
- Verified Deesha certification
Who teaches this
Learn with experienced practitioners
All trainers and mentorsCourses on this pathway are taught by these practitioners. Each batch names its own trainer before you enrol.
Vishal Shah
Founder & Principal Trainer
Two decades building and teaching commerce, banking and cloud platforms — still writing code
- Java & Spring Boot microservices
- TypeScript, React, Angular & Next.js
- Composable commerce (commercetools)
Shrenik Shah
Principal Trainer
Cloud and AI architect who has upskilled over 5,000 engineers in Java, React, Python and cloud
- Python — Django, Flask, GenAI & agentic workflows
- Java & Spring Boot
- React & Angular, micro-frontend architecture
Designed for
- Working developers moving into AI application engineering
- Java and Python backend engineers being asked for AI features
- Final-year and advanced learners with real projects behind them
About this pathway
Every backend developer is about to be asked for an AI feature, and most courses answer with a chatbot notebook. This pathway answers with engineering: the full stack around modern language models — grounding, tooling, evaluation and agents — built inside real applications in the language you already work in.
Why it starts with the assistant
The first two weeks put an agentic coding assistant into your daily workflow, with discipline. That is not a detour: working alongside a strong model every day is the fastest way to develop the judgment the rest of the pathway depends on — what models are reliably good at, where they fail, and how to check.
One spine, two languages
The stack module is a deliberate fork: the same capabilities taught through Python or through Spring AI, chosen once at the start. Everything after the fork — RAG, MCP, evaluation, agents — applies to both.
Who this is for
You need real programming behind you — working Python for the Python branch, working Spring Boot for the Java branch. No AI background is assumed anywhere in the pathway; judgment about AI is the product, not the prerequisite.
Cohorts and fees
Cohorts are small so that every capstone gets individually reviewed. Fees and the next start dates are on this page — or message us on WhatsApp and we will help you choose the entry point that fits.
Reserve a place
Tell us where you are now. We will confirm the intake, the fee and whether this route fits.
Your portfolio when you finish
Design, build, ground, evaluate and deploy AI features and agents inside real applications — in Python or Java — with evidence that they work.
- 01
An AI feature shipped inside a real application
Structured outputs, streaming, tool calling, failure handling and cost controls — a feature users can touch, not a notebook demo.
- 02
A knowledge assistant that cites its sources
RAG over your own documents with measured retrieval quality, honest citations, and a refusal when the evidence is not there.
- 03
An MCP service other tools can depend on
A Model Context Protocol server exposing real capabilities, secured to least privilege and tested with Inspector.
- 04
A tool-using agent with real boundaries
A goal, tools, a budget and a stop condition — plus the written argument for when an agent is the wrong answer.
More pathways
Pick where you want to end up
P-04 · 14–16 weeks
Not sure this is the right route?
Tell us where you are now and we will say plainly whether this pathway fits, or which one does.
Reserve a place