Generative AI, LLMs & AI Agents
From your first LLM API call to building AI systems that can reason, use tools, retrieve information, and perform tasks. Master the engineering foundation powering next-generation autonomous software.
10 Days to Autonomous AI Systems
Go beyond passive chat interfaces. Learn how to architect, ground, and deploy production AI agents that take autonomous actions in software.
Day 1 — LLM Foundations
DAY 01- Transformer mechanics, self-attention, tokenization, and temperature parameters.
- Understanding context window limits, KV caching, and latency tradeoffs.
- Comparing open-source vs. proprietary frontier models (Llama 3, DeepSeek, Claude, GPT-4).
Day 2 — Advanced Prompt Engineering
DAY 02- System prompts, structured role constraints, and few-shot calibration.
- Chain-of-Thought (CoT), Tree-of-Thoughts, and ReAct (Reason + Act) prompting.
- Guarding against hallucinations and enforcing deterministic JSON / Pydantic outputs.
Day 3 — LLM APIs & Integration
DAY 03- Writing robust Python clients for modern LLM provider endpoints.
- Streaming tokens via SSE, handling rate limits, backoff logic, and token cost tracking.
- Structured output generation and schema validation for production codebases.
Day 4 — RAG & Vector Embeddings
DAY 04- Generating dense semantic vector embeddings from custom enterprise documents.
- Chunking strategies: recursive splitting, semantic boundaries, and metadata enrichment.
- Vector database retrieval (ChromaDB / Pinecone), cosine similarity, and grounded context synthesis.
Day 5 — AI Agents Architecture
DAY 05- The difference between simple chains and autonomous agent loops.
- Perception, planning, execution, and evaluation cycles in autonomous agents.
- Designing agent personas, boundaries, and failure-recovery mechanisms.
Day 6 — Tool Calling & External APIs
DAY 06- Function calling mechanisms: schemas, model invocation, and argument parsing.
- Giving LLMs hands: web search tools, database queries, calculators, and API triggers.
- Executing tools safely in sandbox environments with error fallback handling.
Day 7 — n8n + AI Workflow Automation
DAY 07- Connecting modern AI nodes inside visual production workflow orchestrators (n8n).
- Triggering multi-step workflows from Slack, email, webhooks, and forms.
- Combining visual workflow nodes with custom Python and JavaScript agent logic.
Day 8 — Memory & Multi-Step Agents
DAY 08- Short-term buffer memory vs. long-term persistent vector memory stores.
- Conversation state tracking across sessions and multi-turn user intent resolution.
- Orchestrating multi-agent collaboration: supervisor agents, research agents, and critique agents.
Day 9 — Build Your AI Product
DAY 09- Full-day implementation studio: building your original AI product agent.
- Connecting your chosen frontend (FastAPI / Streamlit / Next.js) to the agent backend.
- Optimizing latency, system prompt guardrails, and handling real user queries.
Day 10 — Deployment & Portfolio
DAY 10- Deploying the AI agent backend to production cloud hosting (Render / Fly.io / Vercel).
- Environment variable security, API key management, and production usage observability.
- Project presentation, code architecture walkthrough, and certification graduation.
Possible Capstone Project Domains
Choose one of these industry archetypes or propose your own custom problem to solve:
🎧 Customer Support Agent
Grounds answers in real company knowledge bases, tickets issues, and escalates to human agents when sentiment drops.
🔬 Autonomous Research Agent
Scrapes domain websites, synthesizes findings across 20+ sources, cross-checks facts, and writes executive briefs.
📧 Executive Email Agent
Classifies incoming communication, extracts urgent action items, queries calendar availability, and drafts contextual replies.
📅 Appointment Booking Agent
Conversational booking agent that checks real-time database slots, handles rescheduling, and syncs with Google Calendar.
💼 Inbound Sales Qualifier
Interviews inbound leads via natural chat, scores their budget/need, and logs enriched records into CRM software.
📄 Document Intelligence AI
Parses unstructured invoices, legal contracts, and financial statements into verified, queryable tabular database entries.
Build and Deploy Your Own AI-Powered Application
Graduate with a functioning, production-deployed AI system running in the cloud with live API integrations, vector search grounding, and autonomous decision-making.
Register for Generative AI & AI Agents
Dates: 23–29 November 2026 (Bootcamp). Limited cohort capacity for hands-on code reviews.