
Oct 2026 · LLMs & Agents
Hybrid RAG Search with BM25 and Rank Fusion in Python
Combine exact-term BM25 retrieval and semantic rankings with reciprocal rank fusion. Run a Python baseline and evaluate candidate depth before reranking.
ReadArticles / Topic guide
Topic guide
Resources for moving from AI ideas and prototypes to deployed systems with operational impact.
Built for companies in Colombia, LATAM, and remote teams that need technical implementation, not only strategy.
Related articles

Oct 2026 · LLMs & Agents
Combine exact-term BM25 retrieval and semantic rankings with reciprocal rank fusion. Run a Python baseline and evaluate candidate depth before reranking.
Read
Oct 2026 · LLMs & Agents
Validate extracted invoice amounts against source lines, explicit decimal formats and totals. Measure field accuracy and review coverage with labeled documents.
Read
Sep 2026 · LLMs & Agents
Use the TypeSafe Python SDK and Jev to route requests with typed Choice and Noul results, confidence thresholds and human review.
Read
Sep 2026 · LLMs & Agents
Expose a small, read-only Python service through MCP with validated inputs, bounded results, and an integration checklist.
Read
Sep 2026 · LLMs & Agents
Use a small executable Python harness to measure retrieval and annotated evidence availability before changing a RAG system.
Read
Sep 2026 · LLMs & Agents
Choose what an agent sees from retrieval, memory, and tool results with explicit freshness, provenance, and token budgets.
Read
Jun 2026 · LLMs & Agents
Extract invoices and job postings with Python, Pydantic and Instructor. Validate types, review failures and measure field-level accuracy.
Read
Jun 2026 · LLMs & Agents
Build an autonomous data analysis agent using LangGraph. Learn how agents differ from RAG, and implement the ReAct loop with tools for querying and summarizing data.
Read
Jun 2026 · MLOps
Go from fragile training scripts to robust, observable ML workflows using Prefect. Schedule retraining, handle failures gracefully, and monitor every run.
Read
Jun 2026 · MLOps
Compare batches with Evidently 0.7.23, define numeric features, export drift metrics and tests, and evaluate labeled model quality separately.
Read
May 2026 · Machine Learning
Train a Titanic classifier in Python with fetch_openml, preprocessing, GridSearchCV and a held-out test set. Includes handling unknown categories.
Read
May 2026 · Operations Research
Build an employee roster with pyworkforce 0.5.4: coverage, availability, minimum hours, solver statuses and the limits of its rest and preference rules.
Read
Apr 2026 · Machine Learning
Use sklearn-genetic-opt adapters to schedule mutation and crossover, check their finite-step values, and compare them with a fixed baseline.
Read
Apr 2026 · Machine Learning
Run GAFeatureSelectionCV 0.14.0 with fold-local scaling, inspect the selected features and evaluate baseline and tuned pipelines on a separate holdout.
Read
Apr 2026 · Data Systems
Run a local Kafka and Isolation Forest demo in Python with three partitions, event validation, and offset commits after processing succeeds.
Read
Mar 2026 · MLOps
Train an Iris classifier and serve it with FastAPI. Protect both prediction routes, validate finite inputs, bound batches and test a Docker container.
Read
Mar 2026 · Machine Learning
Compare grid search, random search, and evolutionary algorithms for hyperparameter optimization with sklearn-genetic-opt.
Read
Feb 2026 · Machine Learning
Hyperparameter tuning with GASearchCV from sklearn-genetic-opt using evolutionary algorithms.
Read
Feb 2026 · MLOps
Track Python experiments with MLflow 3.14 and SQLite, separate validation from final testing, and retrieve a versioned model through a registry alias.
Read
Jan 2026 · LLMs & Agents
Build a PDF RAG pipeline with Python, LangChain and Chroma. Preserve page references, validate citation IDs, and separate retrieval tests from answer quality.
Read
Jan 2026 · Operations Research
Plan a repeating workweek with pyworkforce 0.5.4. Verify a nine-person coverage plan and compare minimum headcount with minimum absolute staffing gap.
ReadTell me what you are implementing and where the problem lies: retrieval, models, data or product integration.