AI Software Engineering

Term 0

Foundations

Assumes nothing. Gets a complete beginner to 'I can code and ship.'

+ Learning

·How computers, the internet, and software actually work
·Computational thinking and problem decomposition
·Python from zero — variables, control flow, functions, data structures, basic OOP
·Command line and Git / GitHub from day one
·How AI works intuitively — ML, LLMs and prompting fundamentals
·The math you actually use — logic, statistics, probability, linear algebra feel

+ Immersion

Zomato — Restaurant Discovery & Listing

·Understand how Zomato's restaurant listing system works end-to-end
·Build a CLI tool that reads restaurant data and outputs filtered results
·Scrape and format local business data into structured records
·Ship a live portfolio website documenting what you built and why

Search & Filter

Restaurant search engine

· Parse a dataset into a Python dict / list structure
· Filter by cuisine, price range, minimum rating
· Sort by distance using lat / lng approximation
· Clean, readable CLI output

Data Model

What a record contains

· Design a Restaurant class with all required fields
· Write a validator that flags incomplete records
· Serialize to JSON and back — the foundation of every API
· Load 50 real local businesses into the model

Term 1

Programming & the Web

AI-assisted from the start.

+ Learning

·Practical data structures and algorithms — the interview-relevant core
·Web fundamentals — HTML / CSS / JS, HTTP, client-server, APIs
·Frontend with React; backend with FastAPI / Node; SQL + intro NoSQL
·AI-native development with Claude Code, Cursor, Copilot — and knowing when not to trust them
·Collaboration workflows — branches, pull requests, code review

+ Immersion

Uber — Ride Request & Driver Matching

·Build the full-stack ride-request flow: request → nearest driver
·REST API for ride creation, driver location updates, status polling
·React frontend with map view, fare estimate, live ride status
·First open-source PR merged to an existing project

Ride Request API

The endpoint that starts every trip

· POST /rides with pickup, drop, user_id
· Fare estimate via Haversine distance × rate
· Return ride_id, estimated_fare, pickup time
· Persist to Postgres with status = requested

Driver Matching

Finding the nearest driver in real time

· Drivers table with current lat / lng and status
· Spatial query for the 5 nearest available drivers
· Assign the closest; update statuses atomically
· GET /rides/:id for the frontend to poll every 3s

Live Map Frontend

The map view every rider uses

· React app with Google Maps or Leaflet.js
· Pickup / dropoff markers and route polyline
· Poll every 3s and animate the driver marker
· Show driver info, fare estimate, ETA countdown

Surge Pricing

How Uber 2x's at 2am

· Count active requests in a geographic grid cell
· Count available drivers in the same cell
· Compute requests / drivers, clamp 1.0 — 3.0
· Apply multiplier and surface 'surge active' in UI

Term 2

AI Application Development

+ Learning

·LLMs in depth — tokens, context, embeddings; structured outputs
·Building with LLM APIs — OpenAI, Anthropic, open-source models
·RAG — chunking, embeddings, vector databases, knowledge assistants
·Tool use and function calling; chatbots and assistants; voice and multimodal
·Evaluating AI apps; cost and latency awareness; responsible AI basics
·ML concepts — training vs. inference; fine-tune vs. prompt vs. retrieve

+ Immersion

Swiggy + Zomato — AI-Powered Food Discovery

·Add an AI layer to the food product from Term 1 — natural language search, recommendations, support
·Build a RAG pipeline over a restaurant and menu database
·Ship a real AI app with documented users and usage metrics
·Run a Lamahatta offsite sprint — ship the next major feature in 48 hours

Natural Language Search

Ask for food the way you think

· Embed restaurants and dishes with a text embedding model
· Store vectors in Pinecone, Weaviate or pgvector
· Cosine similarity search on user queries
· Rerank by rating × relevance before display

Recommendation Engine

The 'For You' feed

· Store order history and ratings per user
· Prompt the LLM with history + available menu
· Return a ranked list of 5 dishes per visit
· A/B test vs. random; measure click-through

AI Customer Support

Handle 80% of queries

· RAG over Swiggy-style FAQs
· Tool use: get_order_status(order_id)
· Cover the top 5 query types end-to-end
· Measure containment rate — % resolved without escalation

Menu Intelligence

Reads a menu, extracts data

· Accept a menu image or PDF
· GPT-4o vision extracts dish, price, tags
· Output structured JSON for the database
· Flag low-confidence rows for human review

Term 3

Agents & Systems

Advanced AI engineering.

+ Learning

·AI agents — planning, memory, tool use, multi-agent systems
·Building autonomous workflows and automations
·Advanced RAG, hybrid search, evaluation pipelines
·System design fundamentals — designing scalable AI systems
·Cloud deployment — APIs, webhooks, queues, serverless
·Responsible AI and security — prompt injection, privacy, safety

+ Immersion

WhatsApp — Messaging, Media & AI Agents

·Build a WhatsApp-like messaging system with an AI agent layer
·Real-time message delivery, media handling, read receipts
·Ship an open-source AI contribution merged by external maintainers
·Big-hub trip — Bangalore or Kolkata company visits, hackathon, live demos

Real-Time Messaging

Delivering in under 100ms

· WebSocket connections via Socket.io
· Sent → delivered → read receipt states
· Postgres-backed conversation storage
· Offline queue, flush on reconnect

Media Upload & Preview

Sending without burning data

· Multipart upload with type and size validation
· Server-side compression and thumbnails
· Originals in S3 / R2 with signed URLs
· Range-request video streaming

AI Business Bot

Autonomous customer agent

· ReAct agent with tools — order status, refund, address, escalate
· Full loop — message → tool → response
· Conversation memory across the thread
· Confidence-based escalation to a human queue

E2E Encryption Basics

Why your messages are private

· Public / private key pair via Web Crypto
· Encrypt with recipient's public key
· Decrypt with recipient's private key
· Key exchange handshake for new conversations

Term 4

Production, Scale & Specialisation

Pick a track — AI Product Engineering · Applied ML · AI Infrastructure / DevOps · Full-stack AI.

+ Learning

·Production engineering — CI / CD, testing, monitoring, observability
·MLOps / LLMOps; deployment at scale; cost optimisation
·Track-specific depth — fine-tuning, model deployment, RAG at scale, infrastructure
·Larger open-source contribution and maintainership

+ Immersion

LinkedIn + YouTube — Feed, Ranking & Creator Monetisation

·Build the systems that decide what a million users see — and how creators get paid
·Personal capstone — a production-grade product live in the market
·Lead a junior pod — mentor Term 0 and Term 1 students
·Demo Day prep — public case study, live metrics, investor-ready pitch

Feed Ranking

Why some posts hit 100k views

· Score = recency × engagement × connection distance
· Candidate generation from 1st and 2nd connections
· Top 20 ranked feed
· A/B two formulas; measure dwell time

Video Pipeline

Upload → 'video is live'

· Raw upload to object storage
· Background transcode to 360p / 720p / 1080p with FFmpeg
· Auto-generate 3 thumbnails for creator choice
· State machine — uploading → processing → ready → published

Creator Revenue

How creators see today's earnings

· Track watch duration, ad impressions, ad clicks
· Estimate revenue per video from CPM
· Dashboard — views, watch hours, top earners
· 30-day projection from last 7 days

AI Profile Optimiser

More recruiter views

· Score profile sections with an LLM
· 3 specific actionable suggestions per section
· Rewrite headlines and summaries
· Before / after comparison view

Launchpad

Apprenticeship & Placement

Every student exits with a concrete next step — not a certificate and a hope.

+ Learning

·Paid internship, remote contract, or your own startup with school support
·Job search executed — applications, interviews, freelancing pipeline live
·Founders' track — incubation, first customers, optional revenue-share or school fund
·Demo Day — presented to local and virtual audiences, employers, and investors

+ Immersion

Graduation = Proof

·A portfolio of features from products the world already uses
·Live GitHub repos, deployed apps, and real metrics
·A network of mentors, founders, and hiring managers
·A concrete next step the day you walk out

AI Software
Engineering.

The Terms.

Foundations

+ Learning

+ Immersion

Programming & the Web

+ Learning

+ Immersion

AI Application Development

+ Learning

+ Immersion

Agents & Systems

+ Learning

+ Immersion

Production, Scale & Specialisation

+ Learning

+ Immersion

Apprenticeship & Placement

+ Learning

+ Immersion

Exit with a job, a contract, or a company.

Ready to build?

AI SoftwareEngineering.

The Terms.

Foundations

+ Learning

+ Immersion

Programming & the Web

+ Learning

+ Immersion

AI Application Development

+ Learning

+ Immersion

Agents & Systems

+ Learning

+ Immersion

Production, Scale & Specialisation

+ Learning

+ Immersion

Apprenticeship & Placement

+ Learning

+ Immersion

Exit with a job, a contract, or a company.

Ready to build?

AI Software
Engineering.