Tools and methods · Özer Cem KELAHMET

AI and machine learning

Learning paradigms

Supervised learning Unsupervised learning Semi-supervised learning Reinforcement learning Self-supervised learning Contrastive learning Federated learning Transfer learning Domain adaptation Few-shot learning Zero-shot learning Meta-learning Curriculum learning Active learning

Reinforcement learning

DQN Double DQN Dueling DQN Rainbow REINFORCE A2C A3C PPO TRPO SAC TD3 DDPG Multi-agent RL

Self-supervised methods

SimCLR MoCo BYOL DINO MAE

Neuromorphic

Intel Loihi IBM TrueNorth SpiNNaker BrainScaleS Lava snnTorch Norse SpikingJelly

Foundations

Probability Statistics Linear algebra Optimization Information theory Variational inference Bayesian deep learning Causal inference

LLMs, retrieval, and evaluation

Frontier models

Claude Fable 5.1 Claude Opus 5.5 Claude Sonnet 5.5 Claude Haiku 5.5 GPT-6 GPT-5.6 Gemini 4 Gemini 3.8 Flash Grok 4.7 Mistral Large 4

Open-weight models

DeepSeek V4.1 Qwen 4 Qwen3.8 Kimi K3 GLM-5.3 Gemma 4

Fine-tuning

Full fine-tuning Continued pre-training SFT LoRA QLoRA DoRA rsLoRA GaLore Adapters Prefix tuning Prompt tuning IA3 BitFit

Alignment

RLHF DPO IPO KTO ORPO SimPO RLAIF Constitutional AI

Training tooling

PEFT TRL Axolotl Unsloth LLaMA-Factory Torchtune DeepSpeed ZeRO FSDP Megatron-LM Colossal-AI FlashAttention

Quantization

GPTQ AWQ GGUF bitsandbytes SmoothQuant ExLlamaV2 TurboQuant

RAG

Naive RAG Advanced RAG Modular RAG Agentic RAG GraphRAG Contextual Retrieval HyDE Query rewriting Hybrid search BM25 Cross-encoder reranking ColBERT RAGAS TruLens DeepEval

Vector search

PineconeWeaviateChromaDBMilvusQdrantVespaFAISSpgvectorHNSWIVF-PQScaNNDiskANN

Evaluation and safety

LangSmithLangfusePhoenixBraintrustPromptFoolm-evaluation-harnessGuardrails AINeMo GuardrailsPresidioRed-teaming

Agentic AI

Agent SDKs

Claude Agent SDKOpenAI Agents SDKGoogle ADKMicrosoft Agent FrameworkLangGraphCrewAIPydantic AIMastrasmolagentsLlamaIndexDSPy

Protocols

MCPA2AFunction callingStructured outputsOAuth for tool access

Agent patterns

ReAct loopPlan-and-executeReflectionTree-of-thoughtsOrchestrator-workersMulti-agent systemsHuman-in-the-loopLLM-as-judgeGuardrails

Agent systems

Claude CodeComputer useBrowser agentsLong-running tasksAgent memorySkillsSub-agents

Workflow automation

n8nTemporalCamundaApache AirflowUiPath

Agent reliability

TracingEvalsRegression boardsCost and latency budgetsSandboxingPrompt-injection defense

Foundation models and applied ML

Foundation models from scratch

Architecture design Tokenizer training BPE SentencePiece Synthetic data generation Deduplication Curriculum scheduling Data parallelism Tensor parallelism Pipeline parallelism Mixed precision Gradient checkpointing

Tabular foundation models

Axial attention Perceiver IO Masked cell modeling Imputation networks Semantic sector heads

Related models and baselines

TabPFN TabICL TabSTAR FT-Transformer SAINT TabNet XTab

Continual learning

EWC MIRAS Synaptic Intelligence LwF Experience replay

Self-supervised for tables

Denoising autoencoders InfoNCE MIDAS

Classical ML

scikit-learn XGBoost LightGBM CatBoost Random forest SVM Isolation forest One-class SVM PyCaret SHAP LIME Optuna Ray Tune Bayesian optimization

Forecasting and tasks

ARIMA Prophet N-BEATS TFT PatchTST Chronos Anomaly detection Learning to rank Recommendation Clustering UMAP

Vision and multimodal

OpenCV YOLO Mask R-CNN DETR SAM CLIP BLIP-2 LLaVA Stable Diffusion ControlNet Whisper NeRF Gaussian splatting GIS

Data engineering and lakehouse

AWS data analytics

Amazon S3 AWS Glue Data Catalog AWS Lake Formation Amazon EMR EMR on EKS EMR Serverless Amazon Athena AWS DataSync Amazon MSK

Table formats and lakehouse

Apache Iceberg Iceberg cataloging Iceberg compaction Schema evolution Medallion architecture Bronze layer Silver layer Gold layer Delta Lake

Batch and streaming processing

Apache Spark PySpark Spark Streaming Apache FlinkApache Airflowdbt Trino

Event streaming

Apache Kafka Kafka Connect Single message transforms Schema Registry MirrorMaker 2 MSK Replicator Event-driven architecture RabbitMQ NATS

Data protection

Data tokenization Format-preserving encryption Protegrity Thales Spark UDF de-identification HashiCorp Vault

Databases and sync

Amazon RDS PostgreSQL Oracle Database Change data capture Data synchronization Snowflake Databricks ClickHouse MongoDB Cassandra Redis Elasticsearch

DataOps

GitLab CI/CD Terraform AWS CDK Schema-drift detection Data contract validation Infrastructure as code

Governance

Informatica Axon Informatica EDC Data catalog Data lineage

Engineering and infrastructure

Languages

PythonGoC#C++ScalaTypeScriptRustNode.jsRubySolidity

Backend

FastAPIFlaskDjangoASP.NET.NET CoreAkkaSQL ServerSAP HybrisSOAP

Frontend

ReactNext.jsReact NativeAngularVue.jsSvelteFlutter

Architecture

MicroservicesDomain-driven designCQRSEvent sourcingHexagonal architectureSaga patternCircuit breaker

APIs and security

RESTgRPCGraphQLOpenAPIOAuth 2.0OIDCJWTmTLSZero TrustOWASPGDPRKVKK

ML frameworks

PyTorchPyTorch LightningTensorFlowJAXKerasHugging Face

MLOps and serving

MLflowKubeflowWeights & BiasesBentoMLKServeRay ServeTritonvLLMSGLangTensorRTONNX Runtimellama.cpp

Cloud and GPU

AWSAzureGCPSageMakerVertex AIKubernetesNVIDIA H100NVIDIA H200NVIDIA B200CUDANCCLCloud TPU

DevOps and observability

DockerHelmIstioArgoCDGitHub ActionsJenkinsPrometheusGrafanaOpenTelemetryELK stack

Industrial IoT

OPC UAModbusProfibusMQTTSCADAMESDigital twinsPredictive maintenanceSDN/NFVNetwork slicing

Testing

pytestJestPlaywrightk6

Domains

Industries

Smart manufacturingIndustry 4.0Smart cityUrban intelligenceTelecomFinTechLogisticsE-commerceOnline travelReal-time systemsAugmented reality