Experienced real-time trainers
Learn with trainer-led explanations, demonstrations and guided practice.
Home / Courses / Guntur
Guntur centre · Instructor-led learning
Agentic AI Course in Guntur
Build agents that can act, explain and recover.
Learn Agentic AI in Guntur through practical Python agent builds. Connect tools and knowledge, manage workflow state, add human approvals and evaluate behaviour through guided labs and a portfolio capstone.
PythonLLM APIsLangGraphLangChainPydanticRAGVector SearchMCP ConceptsFastAPIGit
Nipuna experience21+ years
Curriculum18 modules
PracticeGuided labs & projects
Learning supportLMS where applicable
GUIDED PRACTICAL LEARNING
Learn through logic, practice, testing and instructor guidance
Connect each course concept to a practical task, test the result and explain why the approach works.
Plan the logicPractise the skillTest edge casesExplain the result
21 years of coaching and training experience since 4 May 2005
The Nipuna learning support around this course
Exact availability is confirmed for the selected course, Guntur branch and current batch.
Class recordings where applicable
Ask the selected branch about recordings available for the current course and batch.
Nipuna LMS access
Use the learning platform and batch resources provided for the enrolled programme.
Practical labs, assignments and projects
Apply course concepts through the work planned for the selected programme and learner level.
Nipuna AI/LLM Career Guidance
Use structured guidance to compare roles, skills and practical next steps.
Career and placement assistance
Receive resume, interview and job-search support where applicable; employment is not guaranteed.
COURSE OVERVIEW
Build skills you can demonstrate
Move from a model that answers questions to an application that can choose validated tools, retrieve evidence and track a task. The course develops programming and evaluation skills for bounded agents; it does not present unrestricted autonomy as a production shortcut.
WHO CAN JOIN
Choose a course that fits your starting point
Discuss your background and learning goal before choosing a batch.
01
Python learners and graduates
Suitable if you can work with functions, dictionaries and basic APIs. Ask for a readiness discussion if you are starting programming.
02
Software and AI developers
Extend application development into tools, agent state, retrieval, evaluations and controlled deployment.
03
Technical professionals
Build a practical portfolio around a specific business problem with measurable behaviour and documented limitations.
CHOOSE YOUR DIRECTION
Agentic AI, automation and your next step
The two new courses have distinct goals. Choose the depth and working style that fit your needs.
01
Agentic AI
A code-focused course in bounded agents, orchestration, tools, memory, RAG and evaluation. Basic Python knowledge is recommended.
02
AI Automation
A workflow-focused course in n8n, Make, Zapier, APIs and business integrations. Choose this if your priority is automating recurring operations.
03
Generative AI Developer with Python
A broader existing programme that includes Python and GenAI application foundations. Admissions can help you choose a suitable starting point.
PRACTICAL LEARNING
A clear route from concept to working project
Define
Set the task, tools, limits and success criteria.
Build
Connect models, state, retrieval and validated tools.
Review
Add human approval where actions need oversight.
Evaluate
Test success, failure, grounding, cost and recovery.
TOOLS & CONCEPTS
Practise the tools behind the workflow
Exact platforms, libraries, connectors and model providers may vary by trainer and batch. Confirm paid account and API-credit arrangements before enrolment.
01
Python
Build reusable agent functions, handle API responses and test behaviour.
02
LLM APIs
Connect models, structure responses and handle timeouts, errors and usage costs.
03
LangGraph
Model stateful agent workflows with checkpoints and human review.
04
LangChain
Connect retrieval, model and tool components in guided applications.
05
Pydantic
Validate structured inputs and outputs before a tool can act.
06
RAG
Retrieve relevant documents and ground answers in supporting evidence.
07
Vector Search
Index and retrieve relevant content, then measure retrieval quality.
08
MCP Concepts
Understand structured tool connections, permissions and trust boundaries.
09
FastAPI
Expose a tested application through an API with validation and clear responses.
10
Git
Track changes, organise project work and document a reproducible handover.
GUIDED PRACTICE
24 practical lab tasks
Use approved test data, protect credentials, check normal and failure paths, and record execution evidence. The trainer confirms the sequence and practical scope for the current batch.
01
Practice tasks 01–05
Build, test and record evidence of the result.
- Compare a deterministic workflow and an agent on the same task
- Prepare a Python environment and protect model credentials
- Validate a structured model response against a schema
- Write and compare versioned agent instructions
- Create a read-only tool with argument validation
02
Practice tasks 06–10
Build, test and record evidence of the result.
- Build a bounded tool-calling loop
- Add timeouts, iteration limits and failure responses
- Represent a workflow as LangGraph state and transitions
- Resume a checkpointed task after interruption
- Isolate session memory between test users
03
Practice tasks 11–15
Build, test and record evidence of the result.
- Ingest and chunk approved reference documents
- Retrieve evidence and return source citations
- Evaluate missing or conflicting retrieval results
- Build a retrieval-routing agent
- Inspect an MCP tool connection and its permission scope
04
Practice tasks 16–20
Build, test and record evidence of the result.
- Pause an external action for human approval
- Reject an action and resume with corrected instructions
- Compare one-agent and specialist-handoff designs
- Create an agent regression evaluation set
- Test prompt injection with synthetic documents
05
Practice tasks 21–24
Build, test and record evidence of the result.
- Redact sensitive fields from traces and logs
- Measure task cost, latency and tool-call counts
- Expose an agent through an authenticated test API
- Demonstrate the capstone, failure case and recovery runbook
BUILD YOUR PORTFOLIO
Project briefs with a clear purpose
Project scope is adapted to learner readiness. Keep requirements, tests, limitations, review notes and a demonstration of your contribution.
01
Grounded knowledge agent
Answer from approved documents with citations, retrieval checks and an explicit no-answer path.
02
Research and evidence assistant
Use permitted sources, track evidence and produce a reviewable brief with clear uncertainty.
03
Support triage agent
Classify synthetic requests, retrieve relevant guidance and propose actions for staff approval.
04
Document review assistant
Extract structured information, validate fields and route uncertain cases to a reviewer.
05
Approved task execution agent
Plan an action, validate tool arguments, obtain approval and record the result without duplicate writes.
06
Agentic AI capstone
Deliver a selected agent use case with an evaluation set, security review, cost notes, demo and runbook.
LEARNING OUTCOMES
What you should be able to demonstrate
Outcomes depend on attendance, practice and completion of assigned work.
01
Design an appropriate agent
Explain when an agent is useful and when a deterministic workflow is enough.
02
Connect tools and evidence
Build validated tool calls, retrieval and useful task state.
03
Evaluate behaviour
Measure task completion, correctness, grounding, latency and cost.
04
Present a credible portfolio
Demonstrate a working use case, failure handling and documented limitations.
CAREER GUIDANCE
Turn practical work into a credible portfolio
Relevant portfolio directions include AI application development, junior agent engineering and AI integration work. Suitability depends on prior skills, practice and employer requirements. Career assistance does not guarantee employment, salary or placement.
Ask about applicable Nipuna Careers AI/LLM guidance, resume support, interview preparation and portfolio review for your selected branch and batch.
GUNTUR COURSE GUIDANCE
Discuss Agentic AI training in Guntur
For learners in Guntur, start with a discussion of your Python readiness and the agent projects you want to build. The Guntur admissions team can confirm classroom arrangements, current batch timings and available learning modes.
Guntur Training Centre
Door No. 6-4-35, 1st Floor, 4/1 Arundelpet, Opposite Aditya Grand Hotel, Guntur, Andhra Pradesh 522002
+91 79979 27111
admin@nipunatechnologies.com
Suitable for learners around
ArundelpetGuntur
COURSE CURRICULUM
What you will learn
Modules can be updated, reordered or disabled from the admin panel.
01 Agentic AI foundations and system design
Choose an agent only when adaptive tool use adds value.
- Chatbots, deterministic workflows and agents
- Goals, observations, actions and stop conditions
- Single-agent and multi-agent trade-offs
- Defining scope, success criteria and human ownership
02 Python and API readiness
Prepare the programming skills needed for the agent labs.
- Functions, dictionaries, classes and type hints refresher
- Virtual environments and dependency management
- HTTP requests, JSON and asynchronous task awareness
- Git, environment configuration and secret handling
03 LLM APIs and structured responses
Build model calls with validated application outputs.
- Provider-neutral request and response patterns
- Tokens, context windows and sampling concepts
- Structured JSON and Pydantic validation
- Prompt templates, system instructions and versioning
04 Tool use, schemas and function calling
Enable an LLM to call your own application tools.
- Tool descriptions and argument schemas
- Function-calling protocols and tool selection
- Validating tool arguments before execution
- Handling missing parameters, errors and retries
05 Agent loops and execution control
Build dependable loops that terminate correctly.
- Plan, execute, observe and reflect loops
- Iteration limits, timeouts and runaway prevention
- Early stopping and clear failure responses
- Logging and execution history for debugging
06 LangGraph: state, nodes and edges
Model complex agent workflows as observable graphs.
- Graph concepts: state schemas, nodes and edges
- Conditional routing and branching
- Compiling and running a LangGraph agent
- Inspecting and debugging graph execution
07 LangGraph: persistence, memory and human approval
Build agents with checkpointed state and human oversight.
- Checkpointers and thread persistence
- Short-term and long-term memory patterns
- Interrupting a graph for human review
- Modifying state and resuming execution
08 Document ingestion, chunking and embedding
Prepare content so an agent can retrieve reliable evidence.
- Loading text, PDF and web content
- Chunking strategies, metadata and provenance
- Embedding models and vector representation
- Storing and indexing vectors
09 Retrieval-Augmented Generation for agents
Connect retrieval into the agent decision flow.
- Similarity search, filters and reranking
- Constructing grounded prompts with sources
- Handling missing, ambiguous or conflicting context
- Returning verifiable citations with answers
10 Agentic RAG and self-reflection
Make retrieval dynamic and self-correcting.
- Routing queries across multiple knowledge sources
- Grading retrieval relevance before generating an answer
- Query rewriting and second-attempt retrieval
- Detecting hallucinations and ungrounded statements
11 Model Context Protocol (MCP) concepts
Understand structured and secure tool ecosystems.
- What MCP is and why standard tool protocols matter
- MCP clients, servers, resources and tools
- Connecting an agent to an MCP service
- Permission scoping and authentication in MCP tools
12 Multi-agent architectures and handoffs
Design systems of cooperating specialist agents.
- Supervisor, router and collaborator patterns
- Explicit state handoffs between agents
- Avoiding infinite conversation loops
- Consolidating outputs into a unified answer
13 Human-in-the-loop and action authorization
Enforce safety boundaries on consequential agent actions.
- Read-only versus mutation actions
- Implementing approval checkpoints
- User feedback, overrides and re-planning
- Audit logs and accountability for tool execution
14 Evaluation, benchmarks and quality gates
Measure agent behaviour on consistent evaluation sets.
- Why agents need evaluation sets rather than manual testing
- Task completion, tool accuracy and grounding metrics
- Automated evaluation pipelines and LLM-as-a-judge limits
- Regression testing when updating instructions or models
15 Security, privacy and guardrails
Reduce risks in agent inputs, outputs and connected systems.
- Prompt injection and data-exfiltration scenarios
- PII minimisation and secret redaction
- Sandboxed execution and least-privilege tool access
- Abuse handling, retention and incident escalation
16 Reliability, tracing and cost control
Operate agents with visibility into failures and resource use.
- Structured logs, traces and correlation IDs
- Timeouts, retry backoff and idempotent writes
- Budget, latency and concurrency limits
- Fallback providers and operational alerts
17 FastAPI delivery and deployment practice
Expose a reviewed agent service through a controlled application interface.
- Request validation and authentication boundaries
- Background execution and status polling concepts
- Environment separation and deployment configuration
- Health checks, rollback and maintenance runbooks
18 Agentic AI portfolio capstone
Present a bounded agent with evidence of useful and safe behaviour.
- Problem definition and architecture
- Tools, retrieval, approvals and evaluation set
- Security review, failure demonstration and cost notes
- README, demonstration and portfolio explanation
FREQUENTLY ASKED QUESTIONS
Agentic AI questions
What will I learn in the Agentic AI course in Guntur?
You will practise Python-based agents, tool calling, LangGraph workflows, state and memory, grounded retrieval, approval checkpoints, evaluation and deployment concepts through guided work.
Do I need Python before joining?
Basic Python, functions, collections and API knowledge are recommended. The course includes a readiness refresher, not a complete replacement for a foundational Python programme. Discuss your current level before enrolling.
How is Agentic AI different from AI Automation?
Agentic AI focuses on code-based agents and adaptive tool use. AI Automation focuses on connecting business applications and recurring workflows with platforms such as n8n, Make and Zapier.
Is this the same as the Generative AI Developer with Python course?
No. This is a focused agent-engineering course. The existing Generative AI Developer with Python programme covers broader application foundations. Admissions can help you compare prerequisites and goals.
Which tools are used?
The toolkit includes Python, LLM APIs, LangGraph, LangChain, validation tools, retrieval components, FastAPI and Git. Exact libraries and provider versions are confirmed for the current batch.
Will I work on practical projects?
The content includes guided agent labs and project briefs covering knowledge, research, support, document review and approved task execution, followed by a selected capstone. Project scope depends on learner readiness and the batch plan.
Are LMS access and class recordings available?
Applicable Nipuna LMS resources and class recordings support practice and revision. Confirm the access period, included materials and recording availability for your selected branch and batch.
What are the fee, duration and batch timings?
Contact the Guntur admissions team for the current fee, duration, timetable, trainer and demo availability.
Are model credits or cloud subscriptions included?
Model API usage, cloud hosting and paid platform plans may involve separate charges. Ask what the current batch provides before creating paid accounts or enrolling.
Where are classes held in Guntur?
Use this page's location section for the current Guntur centre address and Directions link. Classroom, live online or hybrid options depend on the current course and batch.
Does the programme include career guidance?
Applicable Nipuna Careers AI/LLM guidance, portfolio discussion and interview support can help you plan next steps. Employment, placement and salary are not guaranteed.
GUNTUR ADMISSIONS
Discuss the right batch before enrolling
Ask about the current trainer, timetable, mode, fee and demo availability.