Course Objective & Scope
When users search product catalogues, internal knowledge bases or regulatory archives without knowing exact keywords, lexical search (BM25) fails—yet pure dense vector search often misses exact part numbers, acronyms and legal citations. This course covers dense semantic retrieval with bi-encoders (Sentence-Transformers, BGE-M3, E5), lexical and learned sparse retrieval (BM25, SPLADE), hybrid fusion via Reciprocal Rank Fusion (RRF), and second-stage reranking with Cross-Encoders and late-interaction (ColBERT) models.
Document chunking, HNSW indexing and filtered search in Qdrant and PostgreSQL pgvector, fine-tuning domain embeddings with MultipleNegativesRankingLoss and hard negatives, and evaluation with Recall@k, MRR and NDCG@k are practised hands-on.
Who Should Attend?
- Data scientists and ML engineers building enterprise search, e-commerce search and RAG retrieval layers
- Software engineers developing knowledge portals, support deflection and semantic deduplication/matching services
- Teams upgrading Elasticsearch/OpenSearch or PostgreSQL search with neural retrieval
- Prerequisite: Python and basic NLP / database familiarity
Key Takeaways
- Benchmark and select multilingual and Turkish-capable bi-encoder models (BGE-M3, multilingual-E5) for your domain
- Combine BM25 lexical search with dense embeddings using Reciprocal Rank Fusion (RRF)
- Configure HNSW indices, quantisation and payload filtering in Qdrant and PostgreSQL pgvector
- Boost top-k precision significantly by adding a Cross-Encoder reranking stage
- Fine-tune embedding models on domain query-document pairs with hard-negative mining
Curriculum
Lexical vs Semantic Search and Bi-Encoder Architectures
- Strengths and limits of BM25 and the vocabulary mismatch problem
- Sentence-Transformers, BGE-M3, E5 and Matryoshka variable-dimension embeddings
Vector Indexing: Qdrant, pgvector and Chunking
- Context-preserving document chunking and cosine/dot-product geometry
- HNSW and IVF indexing, scalar/binary quantisation and filtered vector search
Hybrid Search and Cross-Encoder Reranking
- Fusing dense and sparse (BM25/SPLADE) retrieval with Reciprocal Rank Fusion (RRF)
- Bi-encoders vs Cross-Encoders and ColBERT in a two-stage retrieval pipeline
Domain Fine-Tuning of Embedding Models
- Building training pairs from search logs and synthetic query generation
- Fine-tuning with MultipleNegativesRankingLoss and hard-negative mining
Retrieval Evaluation and Semantic Language Applications
- Constructing golden test sets and measuring Recall@k, MRR@10 and NDCG@10
- Semantic deduplication, automated FAQ matching and query understanding
Other Courses in This Category
-
Deep Learning with PyTorch
Comprehensive deep learning training with PyTorch 2.x covering tensor operations, autograd, CNNs, Transformers, LLM fine-tuning (LoRA/QLoRA), diffusion models, GNNs, and production deployment.
- Live Course
- 12 modules
-
Neural networks, forward pass, backpropagation, activation and loss functions, AdamW, dropout and batch normalisation, giving engineers the skills to build and train their first deep learning models in PyTorch.
- Live Course
- 5 modules
-
Stable, large-scale deep learning covering BF16/FP16 mixed precision, multi-GPU training with DDP/FSDP, self-supervised and contrastive pretraining, metric learning, knowledge distillation and GPU profiling.
- Live Course
- 5 modules
-
From ResNet and ConvNeXt to Vision Transformers (ViT), Swin, hybrid backbones, State Space Models (Mamba), Mixture-of-Experts (MoE) and multimodal fusion: selecting and designing architectures for your problem and hardware.
- Live Course
- 5 modules
-
Self-attention mechanics, Multi-Head and Grouped-Query Attention, RoPE positional encodings, encoder/decoder/encoder-decoder families, KV caching and FlashAttention, implemented from scratch in PyTorch and with Hugging Face.
- Live Course
- 5 modules
-
The Keras 3 and TensorFlow 2 ecosystem: Functional/Subclassing APIs, high-throughput tf.data pipelines, custom GradientTape loops, tf.distribute and enterprise deployment via SavedModel, TF Serving and TFLite.
- Live Course
- 5 modules
Get in touch
