AI Engineering

44 programs

AI Engineering Training: LLM, RAG, Agents and Production Systems

Build LLM applications, agent systems and production-ready AI infrastructure from scratch to production using GPT, Claude and Llama.

RAG, fine-tuning, LLMOps, multi-agent orchestration, observability, evaluation — the most comprehensive deep-technical AI program catalog in Türkiye.

Browse programs Request corporate program

44

Programs

3days

Avg. duration

3

Levels

4

Sub-areas

TL;DR

One-line answerAI Engineering training — 35+ deep-technical, production-ready enterprise programs on LLMs, RAG, agents and LLMOps that teach engineers to build production-grade AI systems.

LLM Development & Customization: fine-tuning, PEFT/LoRA, RLHF/DPO, quantization, reasoning models and continued pretraining
RAG & Retrieval: vector databases, embedding models, hybrid search, reranker optimization, GraphRAG
Agent Systems: tool-use, planning, memory, multi-agent orchestration (LangGraph, CrewAI, MCP, Browser Agent)
LLMOps & Production: observability (Langfuse/Phoenix), evaluation harnesses, self-hosting (vLLM/Ollama), context engineering, FastAPI

What you get

Why this category

Production-Ready Code Labs

Not slides — every module ships a working Python repo, test suite and deployment manifest.

Self-Hosted + Cloud Deployment

vLLM, Ollama, Triton for on-prem; AWS Bedrock, Azure OpenAI, GCP Vertex for cloud — applied comparatively.

GPT / Claude / Llama Comparison

For each use-case, comparative selection matrix on capability, cost and data policy.

Security + Evaluation Pipeline

Attack simulation with Llama Guard + Garak + PyRIT, continuous evaluation with LLM-as-judge + golden set.

LLMOps + Observability Stack

Langfuse, Phoenix, Helicone, Weave, Braintrust, LangSmith — vendor-agnostic tracing + cost + drift monitoring.

Cost & Latency Optimization

Prompt caching, batch inference, quantization (GPTQ/AWQ/FP8) for up to 70% token cost reduction scenarios.

Catalog

LLMOps & Production

11 / 44 programs shown

All44 LLM Development & Customization RAG & Retrieval Agent Systems LLMOps & Production

How it works

From assessment to production — AI Engineering delivery

1
1. Maturity Assessment
Team level, current infrastructure, target use-cases and data policy assessed; right program combo recommended.
→
2
2. Curriculum Design
Needs + use-case matching: which sub-group (LLM Dev / RAG / Agents / LLMOps) delivered in which sequence.
→
3
3. Hands-On Lab + Project on Your Data
During the program you build a real use-case end-to-end on your (anonymized) company data.
→
4
4. Production Deploy + Post-Training Mentorship
2 months async post-training mentorship: production deploy, observability stack setup, first regression test pipeline.

Real-world examples

Use cases solved with these programs

Enterprise LLM Chatbot (Banking)

banking

BDDK-compliant, audit-trail-enabled, prompt-injection-guarded customer-facing chatbot.

Document RAG System (Legal / Insurance)

legal

Hybrid search + reranker + citation system over contracts + policies + regulations.

Multi-Agent Operations Bot (Logistics)

logistics

Supervised multi-agent orchestrator for shipment planning, route optimization and customer notifications.

Self-Hosted Turkish LLM (Public / Healthcare)

public

Llama/DeepSeek-based Turkish LLM fine-tuning + vLLM deployment without data leaving premises.

FAQ

AI Engineering — questions answered

Who should attend AI Engineering training?

Designed for backend / full-stack engineers, data and ML engineers, platform/SRE teams and technical leads bringing LLM applications to production. Intermediate Python + REST API + async expertise is expected. Beginner-level program is also available (intro to AI + enterprise prompt engineering).

Which LLM models and frameworks are covered?

Model-agnostic: comparative work on GPT-4o, Claude Sonnet/Opus, Llama 3.x, Mistral, DeepSeek R1. Frameworks covered: LangChain/LangGraph, CrewAI, Claude Agent SDK, LlamaIndex, FastAPI, vLLM, Ollama, Pinecone/Qdrant/Weaviate/pgvector, Langfuse/Phoenix/Helicone.

How comprehensively are production deployment, observability and evaluation covered?

Full LLMOps stack: tracing (Langfuse, Phoenix, LangSmith), metrics (latency, cost, hallucination, drift), guardrails (Llama Guard, Garak, PyRIT), evaluation (golden set + LLM-as-judge + regression pipeline), inference serving (vLLM continuous batching, speculative decoding) and quantization (GPTQ/AWQ/EXL2/GGUF/FP8) covered as dedicated modules.

How much focus is on self-hosted / on-premise LLM scenarios?

Critical for organizations with strict data residency. Includes Ollama + vLLM on-prem deployment, OSS LLM selection (Llama, Mistral, DeepSeek, Turkish LLMs), capacity planning, GPU/TPU sizing, Triton/Dynamo inference layer and PagedAttention internals — 3 dedicated programs: Self-Hosted AI Systems + vLLM Internals + Continued Pretraining.

Are these programs individual or corporate?

All available in both formats. Corporate delivery includes labs on your data, NDA, on-site/hybrid delivery and 2-month post-training mentorship. Individual participants can join public cohorts.

Other categories