Sunt data engineer în Italia, învăț să devin consultant financiar. Ce îmi place la amândouă: să construiesc soluții, să rezolv probleme, să descompun obiective mari și complexe în pași mici și clari. Scriu despre asta în fiecare zi. Stai puțin: poate înveți ceva. Poate învăț și eu.https://narcismiclaus.com/ro/https://narcismiclaus.com/ro/programming/python/60-capstone/https://narcismiclaus.com/ro/programming/python/60-capstone/O privire inapoi la cele 60 de lectii, una inainte la incotro merge Python si resursele care te duc de la nivel intermediar la expert.Fri, 19 Jun 2026 00:00:00 GMTpythoncourse-summarynext-stepshttps://narcismiclaus.com/ro/programming/pyspark/60-spark-cluster-health-check/https://narcismiclaus.com/ro/programming/pyspark/60-spark-cluster-health-check/Checklist-ul capstone: ti se da laptopul, ai timp pana la ora 17 sa-ti dai seama ce e stricat.Thu, 18 Jun 2026 00:00:00 GMTpysparksparkdbahealth-checkcourse-summaryhttps://narcismiclaus.com/ro/programming/python/59-ai-vs-ml/https://narcismiclaus.com/ro/programming/python/59-ai-vs-ml/Decizia care nu exista acum cinci ani: folosesti un model gazduit, faci fine-tune unuia open sau antrenezi propriul model?Tue, 16 Jun 2026 00:00:00 GMTpythonaillmmachine-learningdeploymenthttps://narcismiclaus.com/ro/programming/pyspark/59-adaptive-query-execution/https://narcismiclaus.com/ro/programming/pyspark/59-adaptive-query-execution/Coalescing dinamic al partitiilor, gestionarea skew-ului la runtime si comutarea strategiei de join: configurile de stiut si cazurile in care AQE tot nu te poate ajuta.Mon, 15 Jun 2026 00:00:00 GMTpysparksparkaqeoptimizationperformancehttps://narcismiclaus.com/ro/programming/python/58-transfer-learning/https://narcismiclaus.com/ro/programming/python/58-transfer-learning/Calea realista de la zero la un model deep learning functional in 2026: pleci de la unul pre-antrenat si il fine-tunezi pe datele tale.Fri, 12 Jun 2026 00:00:00 GMTpythondeep-learningtransfer-learninghuggingfacefine-tuninghttps://narcismiclaus.com/ro/programming/pyspark/58-debugging-slow-jobs/https://narcismiclaus.com/ro/programming/pyspark/58-debugging-slow-jobs/Bucla sistematica de a-ti da seama ce nu e in regula cu un job lent: citesti UI-ul, gasesti stage-ul lent, te uiti la skew, GC, volum de shuffle, in ordinea asta.Thu, 11 Jun 2026 00:00:00 GMTpysparksparkdebuggingperformanceproductionhttps://narcismiclaus.com/ro/programming/python/57-training-loop/https://narcismiclaus.com/ro/programming/python/57-training-loop/Cele cinci linii care transforma o retea initializata aleatoriu intr-un model antrenat si contabilitatea care le face de calitate productie.Tue, 09 Jun 2026 00:00:00 GMTpythonpytorchtraining-loopdeep-learninghttps://narcismiclaus.com/ro/programming/pyspark/57-memory-tuning/https://narcismiclaus.com/ro/programming/pyspark/57-memory-tuning/Cele patru config-uri care chiar conteaza, ce inseamna spill, cum sa citesti un stack trace OOM si regula pentru sizing-ul executorilor.Mon, 08 Jun 2026 00:00:00 GMTpysparksparkmemorytuningproductionhttps://narcismiclaus.com/ro/programming/python/56-pytorch/https://narcismiclaus.com/ro/programming/python/56-pytorch/Tensori, autograd, modulul nn si feel-ul Python care a facut PyTorch sa castige.Fri, 05 Jun 2026 00:00:00 GMTpythonpytorchdeep-learningtensorshttps://narcismiclaus.com/ro/programming/pyspark/56-execution-plans/https://narcismiclaus.com/ro/programming/pyspark/56-execution-plans/Cum sa citesti fiecare linie din output-ul .explain(), operatorii care conteaza si pasii optimizatorului care ii produc.Thu, 04 Jun 2026 00:00:00 GMTpysparksparkexplainexecution-plancatalysthttps://narcismiclaus.com/ro/programming/python/55-neural-networks-plain/https://narcismiclaus.com/ro/programming/python/55-neural-networks-plain/Ce este de fapt o retea neuronala, de ce functioneaza backpropagation si unde deep learning bate cu adevarat ML-ul clasic.Tue, 02 Jun 2026 00:00:00 GMTpythondeep-learningneural-networksfundamentalshttps://narcismiclaus.com/ro/programming/pyspark/55-spark-ui/https://narcismiclaus.com/ro/programming/pyspark/55-spark-ui/Un tur ghidat al fiecarui tab, Jobs, Stages, Tasks, SQL, Storage, Executors, si ce iti spune fiecare cand ceva nu merge.Mon, 01 Jun 2026 00:00:00 GMTpysparksparkuidebuggingproductionhttps://narcismiclaus.com/ro/programming/python/54-ml-project/https://narcismiclaus.com/ro/programming/python/54-ml-project/De la CSV brut la model deployat: lectiile Modulului 9 facute palpabile.Fri, 29 May 2026 00:00:00 GMTpythonmachine-learningprojectend-to-endclassificationhttps://narcismiclaus.com/ro/programming/pyspark/54-output-modes-and-sinks/https://narcismiclaus.com/ro/programming/pyspark/54-output-modes-and-sinks/Append vs update vs complete, sink-urile pe care le aduce Spark si escape hatch-ul foreachBatch pentru tot restul.Thu, 28 May 2026 00:00:00 GMTpysparksparkstreamingsinksidempotentforeach-batchhttps://narcismiclaus.com/ro/programming/python/53-hyperparameter-tuning/https://narcismiclaus.com/ro/programming/python/53-hyperparameter-tuning/Cele patru strategii de cautare, cand are sens fiecare si de ce optuna e default-ul in 2026.Tue, 26 May 2026 00:00:00 GMTpythonmachine-learninghyperparametersoptunagridsearchhttps://narcismiclaus.com/ro/programming/pyspark/53-stateful-operations/https://narcismiclaus.com/ro/programming/pyspark/53-stateful-operations/Unde tine Spark Streaming state-ul intre micro-batch-uri, pattern-urile stateful standard si cand sa cobori la mapGroupsWithState.Mon, 25 May 2026 00:00:00 GMTpysparksparkstreamingstatesessionizationhttps://narcismiclaus.com/ro/programming/python/52-linear-models/https://narcismiclaus.com/ro/programming/python/52-linear-models/De ce modelele liniare sunt inca raspunsul corect surprinzator de des si trucurile de regularizare care le fac de productie.Fri, 22 May 2026 00:00:00 GMTpythonmachine-learninglinear-regressionlogistic-regressionregularizationhttps://narcismiclaus.com/ro/programming/pyspark/52-watermarks-event-time/https://narcismiclaus.com/ro/programming/pyspark/52-watermarks-event-time/De ce event time conteaza mai mult decat processing time, ce face de fapt un watermark si exemplul lucrat cu timestamp-uri concrete.Thu, 21 May 2026 00:00:00 GMTpysparksparkstreamingwatermarksevent-timehttps://narcismiclaus.com/ro/programming/python/51-tree-models/https://narcismiclaus.com/ro/programming/python/51-tree-models/De ce trees domina ML-ul tabular, diferentele dintre cele trei mari biblioteci si hyperparametrii care conteaza.Tue, 19 May 2026 00:00:00 GMTpythonmachine-learningrandom-forestxgboostlightgbmgradient-boostinghttps://narcismiclaus.com/ro/programming/pyspark/51-kafka-source/https://narcismiclaus.com/ro/programming/pyspark/51-kafka-source/Cum citeste Spark din Kafka, semantica offset-urilor si intrebarea at-least-once vs exactly-once.Mon, 18 May 2026 00:00:00 GMTpysparksparkkafkastreamingstructured-streaminghttps://narcismiclaus.com/ro/programming/python/50-feature-engineering/https://narcismiclaus.com/ro/programming/python/50-feature-engineering/Transformarile care fac din date brute combustibil pentru model si cele care scurg in tacere informatii din viitor.Fri, 15 May 2026 00:00:00 GMTpythonmachine-learningfeaturespreprocessinghttps://narcismiclaus.com/ro/programming/pyspark/50-structured-streaming-basics/https://narcismiclaus.com/ro/programming/pyspark/50-structured-streaming-basics/Punctele de intrare pentru streaming, semantica trigger-elor si checkpoint-ul de care depinde totul.Thu, 14 May 2026 00:00:00 GMTpysparksparkstreamingstructured-streamingdataframehttps://narcismiclaus.com/ro/programming/python/49-scikit-learn/https://narcismiclaus.com/ro/programming/python/49-scikit-learn/Pattern-ul fit/predict care sustine totul, categoriile de modele si pipeline-urile care fac codul reproductibil.Tue, 12 May 2026 00:00:00 GMTpythonscikit-learnmachine-learningmlhttps://narcismiclaus.com/ro/programming/pyspark/49-why-streaming/https://narcismiclaus.com/ro/programming/pyspark/49-why-streaming/Date marginite vs nemarginite, batch-vs-streaming ca un continuum si de ce DStreams sunt depreciate in favoarea Structured Streaming.Mon, 11 May 2026 00:00:00 GMTpysparksparkstreamingstructured-streamingfundamentalshttps://narcismiclaus.com/ro/programming/python/48-numerical-project/https://narcismiclaus.com/ro/programming/python/48-numerical-project/Ia un dataset public si ruleaza o analiza numerica completa: statistici descriptive, ajustari, teste de ipoteza, grafice.Fri, 08 May 2026 00:00:00 GMTpythonnumpyscipyprojectanalysishttps://narcismiclaus.com/ro/programming/pyspark/48-schema-evolution/https://narcismiclaus.com/ro/programming/pyspark/48-schema-evolution/De ce formatele schema-on-read gestioneaza prost schimbarea, de ce Avro+registry o gestioneaza bine si modul Delta/Iceberg de a le obtine pe ambele.Thu, 07 May 2026 00:00:00 GMTpysparksparkschemaparquetavroevolutionhttps://narcismiclaus.com/ro/programming/python/47-jupyter-and-notebooks/https://narcismiclaus.com/ro/programming/python/47-jupyter-and-notebooks/De ce notebook-urile sunt seducatoare, unde stralucesc si momentul in care ar trebui sa te opresti si sa scrii un script real.Tue, 05 May 2026 00:00:00 GMTpythonjupyternotebooksvscodekernelhttps://narcismiclaus.com/ro/programming/pyspark/47-cloud-storage/https://narcismiclaus.com/ro/programming/pyspark/47-cloud-storage/Avertismentele despre consistenta, problema renumirii si de ce exista direct-write committers.Mon, 04 May 2026 00:00:00 GMTpysparksparks3cloudstoragehadoophttps://narcismiclaus.com/ro/programming/architecture/80-capstone/https://narcismiclaus.com/ro/programming/architecture/80-capstone/Optzeci de lecții de arhitectură de sistem, condensate într-un singur exercițiu de design. Aceeași companie SaaS fictivă, trei scări, trei arhitecturi, și un tur ghidat al ce se schimbă și de ce. Lecția de încheiere a cursului.Fri, 01 May 2026 00:00:00 GMTarchitecturecapstonecourse-summaryhttps://narcismiclaus.com/ro/programming/python/features-learned-too-late/https://narcismiclaus.com/ro/programming/python/features-learned-too-late/Match statements, operatorul walrus, debugging cu f-string-uri, dataclasses și alte funcționalități Python care m-ar fi scutit de ore întregi dacă le-aș fi știut mai devreme.Fri, 01 May 2026 00:00:00 GMTpythontipsmodern-pythonhttps://narcismiclaus.com/ro/programming/pyspark/46-writing-jdbc/https://narcismiclaus.com/ro/programming/pyspark/46-writing-jdbc/Cum scrii output-ul Spark inapoi intr-o baza de date relationala fara s-o doboari, sa rupi tranzactiile sau sa pierzi date la retry.Thu, 30 Apr 2026 00:00:00 GMTpysparksparkjdbcwritetransactionshttps://narcismiclaus.com/ro/programming/python/45-scipy/https://narcismiclaus.com/ro/programming/python/45-scipy/Statistica, optimizare, prelucrare de semnal, matrice rare: biblioteca standard a Python-ului stiintific.Tue, 28 Apr 2026 00:00:00 GMTpythonscipystatisticsoptimizationhttps://narcismiclaus.com/ro/programming/architecture/79-ml-platform/https://narcismiclaus.com/ro/programming/architecture/79-ml-platform/Cele cinci straturi pe care s-au standardizat platformele ML moderne, problema de train-serve skew pentru care a fost inventat feature store-ul și calculul build-versus-buy pentru fiecare strat în 2026.Mon, 27 Apr 2026 00:00:00 GMTarchitectureml-platformfeature-storemlflowservinghttps://narcismiclaus.com/ro/programming/pyspark/45-reading-jdbc/https://narcismiclaus.com/ro/programming/pyspark/45-reading-jdbc/Conectorul de sursa JDBC, trucul partitionColumn si de ce o citire naiva iti omoara baza de date sursa.Mon, 27 Apr 2026 00:00:00 GMTpysparksparkjdbcpostgresmysqlparallel-readhttps://narcismiclaus.com/ro/programming/architecture/78-privacy-compliance/https://narcismiclaus.com/ro/programming/architecture/78-privacy-compliance/Reglementările de privacy ca factori arhitecturali. Dreptul la ștergere, data residency, customer-managed keys și infrastructura de consimțământ și audit pe care cadrele de conformitate le cer.Fri, 24 Apr 2026 00:00:00 GMTarchitecturegdprccpaprivacycomplianceresidencyhttps://narcismiclaus.com/ro/programming/python/44-plotting/https://narcismiclaus.com/ro/programming/python/44-plotting/Trei biblioteci de plotting, trei filozofii si la care apelezi in functie de audienta.Fri, 24 Apr 2026 00:00:00 GMTpythonmatplotlibseabornplotlyvisualizationhttps://narcismiclaus.com/ro/programming/pyspark/44-orc-avro-delta/https://narcismiclaus.com/ro/programming/pyspark/44-orc-avro-delta/Trei familii de formate care nu sunt Parquet, cand este fiecare alegerea corecta si de ce Delta a tot preluat in liniste controlul.Thu, 23 Apr 2026 00:00:00 GMTpysparksparkorcavrodeltafile-formathttps://narcismiclaus.com/ro/programming/architecture/77-security-architecture/https://narcismiclaus.com/ro/programming/architecture/77-security-architecture/Principiile de securitate de care fiecare sistem are nevoie ca arhitectură de portanță. Least privilege, defense in depth, zero trust și controalele IAM și de rețea care transformă principiile în realitate.Wed, 22 Apr 2026 00:00:00 GMTarchitecturesecurityiamleast-privilegehttps://narcismiclaus.com/ro/programming/python/43-numpy/https://narcismiclaus.com/ro/programming/python/43-numpy/Ce este un ndarray, de ce broadcasting-ul schimba felul in care scrii bucle si setul mic de functii care acopera majoritatea cazurilor.Tue, 21 Apr 2026 00:00:00 GMTpythonnumpyarraysbroadcastingvectorizationhttps://narcismiclaus.com/ro/programming/architecture/76-disaster-recovery/https://narcismiclaus.com/ro/programming/architecture/76-disaster-recovery/Ce înseamnă de fapt disaster recovery în practică. Cele patru niveluri DR, RTO și RPO ca butoane de proiectare și disciplina exercițiului care dovedește că planul funcționează.Mon, 20 Apr 2026 00:00:00 GMTarchitecturedisaster-recoveryrtorpobackupshttps://narcismiclaus.com/ro/programming/pyspark/43-parquet/https://narcismiclaus.com/ro/programming/pyspark/43-parquet/Stocarea pe coloane explicata, codecuri de compresie, predicate pushdown si structura row-group care face citirile selective rapide.Mon, 20 Apr 2026 00:00:00 GMTpysparksparkparquetfile-formatcolumnarhttps://narcismiclaus.com/ro/programming/architecture/75-multi-region/https://narcismiclaus.com/ro/programming/architecture/75-multi-region/De ce echipele aleg multi-region (latență, DR, conformitate, capacitate), cele trei forme de deployment, problemele grele (replicare, conflicte, cost) și când să nu te obosești.Fri, 17 Apr 2026 00:00:00 GMTarchitecturemulti-regiongeographylatencyhttps://narcismiclaus.com/ro/programming/python/42-data-engineering-project/https://narcismiclaus.com/ro/programming/python/42-data-engineering-project/De la sursa la destinatie, cu monitorizare, idempotenta si planificare. Lectiile Modulului 7 facute tangibile.Fri, 17 Apr 2026 00:00:00 GMTpythondata-engineeringpipelineprojectend-to-endhttps://narcismiclaus.com/ro/programming/pyspark/42-tungsten/https://narcismiclaus.com/ro/programming/pyspark/42-tungsten/Cum fuzioneaza Spark operatiile in cod compilat, formatul columnar off-heap si de ce Spark cu DataFrame e rapid.Thu, 16 Apr 2026 00:00:00 GMTpysparksparktungstenperformanceinternalshttps://narcismiclaus.com/ro/programming/sql-server/40-health-check/https://narcismiclaus.com/ro/programming/sql-server/40-health-check/Checklist-ul ordonat pe care îl rulez pe orice SQL Server pe care nu l-am mai văzut. Imprimabil, lipibil și lecția de încheiere a cursului.Thu, 16 Apr 2026 00:00:00 GMTsql-serverdbahealth-checkcourse-summaryhttps://narcismiclaus.com/ro/programming/architecture/74-event-driven-architecture/https://narcismiclaus.com/ro/programming/architecture/74-event-driven-architecture/Servicii care vorbesc emițând evenimente, alegerea între choreography și orchestration, pattern-ul saga și setul de unelte din 2026 (Temporal, Step Functions, Camunda, Argo).Wed, 15 Apr 2026 00:00:00 GMTarchitectureevent-drivensagachoreographyorchestrationhttps://narcismiclaus.com/ro/programming/python/41-orchestration/https://narcismiclaus.com/ro/programming/python/41-orchestration/Cand cron nu mai e suficient, ce face efectiv un orchestrator si cei trei contestatari pentru pipeline-urile tale de date.Tue, 14 Apr 2026 00:00:00 GMTpythonairflowprefectdagsterorchestrationhttps://narcismiclaus.com/ro/programming/architecture/73-microservices/https://narcismiclaus.com/ro/programming/architecture/73-microservices/Boom-ul microservices din 2015-2020, contraofensiva de după 2021 și modular monolith ca drum de mijloc. Legea lui Conway, taxa pe sisteme distribuite și cum alegi în funcție de mărimea echipei și profilul de scalare.Mon, 13 Apr 2026 00:00:00 GMTarchitecturemicroservicesmonolithhttps://narcismiclaus.com/ro/programming/pyspark/41-catalyst-optimizer/https://narcismiclaus.com/ro/programming/pyspark/41-catalyst-optimizer/Cum transforma Spark codul tau intr-un query plan, cele patru faze de optimizare si cum sa citesti .explain(True).Mon, 13 Apr 2026 00:00:00 GMTpysparksparkcatalystoptimizerexplainhttps://narcismiclaus.com/ro/finance/59-giorgio-catch-up/https://narcismiclaus.com/ro/finance/59-giorgio-catch-up/Ce dacă ai început târziu? Matematica recuperării, opțiuni realiste și avantajul deducerii fiscale al fondo pensione pentru muncitorii mai în vârstă.Mon, 13 Apr 2026 00:00:00 GMTretirementcatch-upgiorgioworked-example