ZTABS builds AI data pipelines for Lisbon businesses — ETL/ELT, feature engineering, model training, and real-time processing. GDPR compliant data infrastructure powering AI initiatives across Portuguese and European markets.
AI data pipeline development in Lisbon serves the growing number of Portuguese and European businesses that have discovered the hardest part of AI isn't the model — it's the data. Building an AI model is relatively straightforward; building the pipeline that reliably feeds that model with clean, timely, properly governed data is where most AI projects succeed or fail. Lisbon's rapidly growing tech ecosystem, fuelled by attractive tax regimes (NHR programme for tech workers), quality of life, and a growing talent pool, has made the city a European hub for AI and data engineering. Lisbon's AI data pipeline landscape: several factors define the data engineering context in Portugal. GDPR as a design constraint: data pipelines in Lisbon must be GDPR-compliant by design — not GDPR-adapted after building. The CNPD (Comissão Nacional de Proteção de Dados — Portugal's data protection authority): actively enforcing GDPR with significant fines for non-compliance. Data pipeline implications: data minimisation (collecting and processing only what's needed), purpose limitation (data used only for stated purposes), right to deletion (pipelines must support data deletion requests across the entire data lifecycle), consent management (tracking and enforcing consent for data processing), and data lineage (knowing where data came from, how it was transformed, and where it went). AI-specific GDPR: the EU AI Act adding additional requirements for AI data — training data documentation, bias detection, and data quality standards for high-risk AI systems. Portuguese tech ecosystem: Lisbon's tech ecosystem producing significant data engineering demand — startups (Web Summit alumni and Lisbon's growing startup scene generating data-intensive businesses), scale-ups (companies like Talkdesk, OutSystems, and Feedzai demonstrating Portuguese tech capability at scale), enterprise digital transformation (traditional Portuguese businesses — banking, insurance, retail, and manufacturing — digitising operations and generating data volumes that require engineering), and European operations centres (international companies establishing Lisbon as their European data operations centre — attracted by talent, cost, and timezone). Data infrastructure maturity: Lisbon businesses at varying stages of data maturity — some with mature data warehouses (particularly banking and telecom), others with data scattered across spreadsheets, SaaS platforms, and legacy systems. Data pipelines: often needing to accommodate this full spectrum — connecting mature databases alongside CSV exports from legacy systems that haven't been updated since Portugal joined the eurozone. Portuguese language data: AI models serving Portuguese markets needing Portuguese language data — customer support transcripts, product reviews, social media sentiment, and document processing. Portuguese NLP: European Portuguese differing from Brazilian Portuguese in vocabulary, grammar, and expression — models trained on Brazilian Portuguese performing poorly on European Portuguese text. Data pipelines: needing to handle Portuguese text with appropriate NLP preprocessing — tokenisation, lemmatisation, and sentiment analysis calibrated for European Portuguese.
Lisbon's AI data pipeline demand comes from businesses building or scaling AI capabilities: (1) Fintech and banking: Lisbon fintech companies and Portuguese banks. Financial data pipelines: transaction processing (real-time transaction data pipelines — processing millions of transactions daily, enriching with merchant data, categorising, and feeding into fraud detection models. Portuguese banking: Multibanco/MB Way transactions adding Portugal-specific payment data to the pipeline), risk modelling (credit risk and market risk data — aggregating customer financial data, external bureau data, and market indicators into feature stores for risk models. Banco de Portugal: regulatory reporting requiring specific data aggregation and formatting), fraud detection (real-time transaction scoring — streaming pipelines processing transactions in milliseconds, extracting features, and scoring against fraud models. Portuguese payment landscape: Multibanco, MB Way, SEPA, and card transactions each requiring specific fraud feature engineering), and regulatory reporting (EBA, ECB, and Banco de Portugal reporting — data aggregation, validation, and formatting for regulatory submissions. GDPR: ensuring reporting data is properly anonymised where required). (2) eCommerce and retail: Portuguese online retailers and marketplaces. Retail data pipelines: recommendation engines (product recommendation data — user behaviour events (clicks, views, purchases), product attributes, and inventory data feeding collaborative filtering and content-based models. Portuguese eCommerce: smaller catalogue sizes than Amazon requiring different recommendation approaches), demand forecasting (sales prediction — historical sales, seasonal patterns, promotional calendars, and external factors feeding time-series models. Portuguese retail: seasonal patterns influenced by local events — Santos Populares, Christmas, Saldos de Janeiro), customer segmentation (customer data — purchase history, browsing behaviour, and demographic data aggregated into customer profiles for segmentation models. GDPR: customer data processing requiring explicit consent tracking throughout the pipeline), and pricing optimisation (competitive pricing data — scraping competitor prices, monitoring marketplace pricing, and feeding pricing models. Portuguese market: smaller market with specific competitive dynamics requiring Portugal-specific pricing intelligence). (3) Healthcare and pharma: Portuguese health organisations and pharmaceutical companies. Healthcare data pipelines: clinical data (patient data from hospital systems — HL7 FHIR integration, lab results, imaging data, and clinical notes. Portuguese healthcare: SNS (Serviço Nacional de Saúde) integration adding public health data sources. GDPR: health data classified as special category — heightened protection requirements), pharmaceutical R&D (drug discovery and clinical trial data — molecular data, trial results, and adverse event reporting. Portuguese pharma: clinical trial data from Portuguese hospitals feeding into European multi-site analyses), and population health (epidemiological data — aggregating de-identified health data for population health analysis. SNS data: Portuguese public health data requiring specific access agreements and anonymisation), and medical imaging (DICOM imaging data pipelines — processing radiology, pathology, and other imaging data for AI-assisted diagnosis. Portuguese hospitals: integrating with existing PACS systems). (4) Manufacturing and industry: Portuguese manufacturing companies. Industrial data pipelines: IoT sensor data (factory sensor data — temperature, pressure, vibration, and quality measurements streaming from manufacturing equipment. Portuguese industry: automotive (Autoeuropa — VW plant), ceramics, textiles, and cork — each with specific sensor data requirements), predictive maintenance (equipment health monitoring — sensor data feeding failure prediction models. The pipeline: handling high-frequency sensor data (thousands of readings per second), extracting features, and feeding prediction models in real-time), quality control (manufacturing quality data — vision inspection results, measurement data, and defect classification feeding quality prediction models. Portuguese manufacturing: quality data often in mixed formats — structured machine data alongside manual inspection records), and supply chain (supply chain data — supplier performance, logistics tracking, and demand signals feeding supply chain optimisation models. Portuguese supply chain: European supply chain integration with specific Portuguese logistics patterns). (5) Energy: Portuguese energy companies and renewables. Energy data pipelines: renewable energy (solar and wind generation data — weather forecasts, historical generation, and grid demand feeding production prediction models. Portugal: significant renewable energy capacity — solar and wind data pipelines critical for grid management), smart grid (grid sensor data — real-time consumption, generation, and network conditions feeding grid management AI. Portuguese grid: transitioning to smart metering creating new data volumes), and energy trading (market price data — MIBEL (Iberian electricity market) price data, generation forecasts, and consumption predictions feeding trading models. Portuguese energy market: integrated with Spain through MIBEL requiring cross-border data handling).
Our Lisbon AI data pipeline development follows a methodology designed for GDPR-compliant, production-grade data infrastructure: (1) Data assessment (weeks 1-2): understanding the data landscape. Source inventory: cataloguing all data sources — databases (PostgreSQL, MySQL, Oracle, SQL Server), SaaS platforms (Salesforce, HubSpot, Shopify, Google Analytics), files (CSVs, Excel, JSON), APIs (internal and external), and streaming sources (Kafka, IoT sensors, event streams). Portuguese data sources: Portuguese-specific systems — Multibanco/SIBS data, AT (Autoridade Tributária) tax data, Portuguese social security (Segurança Social), and Portuguese banking APIs (PSD2/Open Banking). Data quality: assessing data quality across sources — completeness (missing values), consistency (conflicting data between sources), accuracy (verified against known truths), timeliness (how fresh the data is), and relevance (whether the data actually serves the AI use case). GDPR audit: mapping personal data across all sources — what personal data exists, under what legal basis it's processed, how long it's retained, and whether consent covers the intended AI use. The GDPR audit: determining pipeline design constraints — anonymisation requirements, retention policies, and consent enforcement. AI readiness: assessing whether the data can support the intended AI use — sufficient volume, representative coverage, appropriate labels (for supervised learning), and temporal coverage (enough history for time-series models). (2) Pipeline architecture (weeks 2-3): designing the data infrastructure. Architecture selection: choosing the right pipeline approach. Batch processing (scheduled data processing — hourly, daily, or weekly. Best for: analytics, reporting, and model training where real-time isn't required. Technologies: Apache Spark, dbt, Airflow), stream processing (real-time data processing — events processed as they occur. Best for: fraud detection, real-time recommendations, and operational monitoring. Technologies: Apache Kafka, Apache Flink, AWS Kinesis), and hybrid (batch for historical analysis and model training, streaming for real-time serving — the most common pattern for AI data pipelines). Data platform: selecting the infrastructure. Cloud (our default for Lisbon businesses — AWS eu-south-2 (Spain, nearest to Lisbon), Azure West Europe, or GCP europe-west1. GDPR: EU-region deployment ensuring data remains within the EU), on-premises (for highly regulated businesses requiring on-site data processing — Portuguese banks and healthcare organisations with specific data residency requirements), and hybrid (sensitive data on-premises, processed results in cloud — balancing security with scalability). Data modelling: designing the data structures — raw layer (data landed as-is from sources), staging layer (cleaned, validated, and standardised data), curated layer (business-ready data, modelled for analysis), and feature store (ML-ready features, computed and cached for model training and serving). GDPR design: privacy by design — PII identified and tagged, access controls at the column level, anonymisation applied in the appropriate layer, and deletion capability throughout the pipeline. (3) Pipeline development (weeks 3-5): building the data infrastructure. Ingestion: building data connectors for each source. Database connectors (CDC — Change Data Capture — for real-time database synchronisation. Debezium for PostgreSQL/MySQL, Oracle GoldenGate for Oracle), API connectors (scheduled API calls for SaaS platforms — handling pagination, rate limiting, and error recovery), file ingestion (SFTP, S3, and cloud storage monitoring for file-based data sources), and streaming (Kafka producers for real-time event capture — application events, IoT sensors, and transactional systems). Transformation: data cleaning and preparation. dbt (our default transformation tool — SQL-based transformations with testing, documentation, and lineage. dbt: enabling analysts and data engineers to collaborate on transformation logic), PySpark (for complex transformations requiring Python — ML feature engineering, NLP preprocessing, and custom business logic), and Great Expectations (data quality testing — validating data at each pipeline stage, catching issues before they affect downstream models). Portuguese data transformations: NIF (Número de Identificação Fiscal) validation, Portuguese address standardisation, European Portuguese NLP preprocessing, and Portuguese date/number formatting (DD/MM/YYYY, comma as decimal separator). Feature engineering: computing ML features from raw data. Feature store (Feast or custom — computing, storing, and serving ML features. Online store: for real-time model serving. Offline store: for model training), feature computation (aggregations, ratios, time-series features, text embeddings, and categorical encodings — computed from curated data and stored in the feature store), and feature versioning (tracking feature definitions, enabling reproducibility and feature reuse across models). (4) GDPR compliance layer (weeks 4-5): embedding privacy into the pipeline. Data classification: automated PII detection and classification — names, email addresses, phone numbers, NIFs, and other personal data tagged throughout the pipeline. Access control: column-level security — PII accessible only to authorised roles. Anonymised views: for analytics and ML that don't require personal data. Anonymisation: k-anonymity, l-diversity, or differential privacy applied where data is used for analytics or ML training. Portuguese GDPR: ensuring anonymisation meets CNPD expectations — true anonymisation, not pseudonymisation disguised as anonymisation. Right to deletion: data deletion capability across the entire pipeline — from raw storage through transformed layers to feature stores and model training data. Deletion verification: confirming data removal across all pipeline stages. Consent tracking: consent metadata flowing through the pipeline — data processing restricted to consented purposes. Consent withdrawal: triggering automatic data restriction across the pipeline. Audit logging: comprehensive logging of all data processing — who accessed what, when, and for what purpose. GDPR accountability: logs enabling demonstration of compliance to CNPD. (5) Monitoring and operations (weeks 5-6+): keeping the pipeline healthy. Pipeline monitoring: real-time tracking of pipeline health — job success/failure, processing latency, data volume, and error rates. Alerting: automated alerts for pipeline failures, data quality issues, and anomalies. Data quality monitoring: continuous data quality checks — schema drift detection, distribution drift detection, and freshness monitoring. Model drift: tracking input data distribution changes that might affect model performance — alerting data science teams when feature distributions shift. Cost management: cloud cost monitoring — tracking compute and storage costs, identifying optimisation opportunities, and right-sizing infrastructure. Portuguese cost sensitivity: Lisbon businesses often operating on tighter budgets than London or San Francisco — cost efficiency critical in pipeline design.
Fintech / Financial Services
A Lisbon-based fintech company (€12M annual revenue — providing payment processing and financial analytics for Portuguese SMEs, 95 employees, processing €2.8B in annual transaction volume, Banco de Portugal regulated) needed to build AI data pipelines for fraud detection and business analytics. (1) Data chaos: transaction data in PostgreSQL (primary database), merchant data in Salesforce, settlement data in a legacy Oracle system, risk data in spreadsheets, and customer support data in Zendesk. The systems: not connected — analysing cross-system patterns (e.g., correlating transaction patterns with customer support complaints to identify fraud) requiring manual data extraction and Excel merging. Analysis time: a fraud analyst needing 4-6 hours to compile a cross-system view for a single investigation. (2) Fraud detection: the company's fraud detection was rule-based — 147 rules checking transaction attributes (amount, velocity, merchant category, and geographic anomaly). Rule maintenance: rules added reactively after fraud incidents — never removed, often conflicting, and impossible to tune holistically. False positive rate: 8.2% — 8.2% of legitimate transactions flagged as potentially fraudulent, requiring manual review. Manual review team: 6 analysts reviewing 4,100 flagged transactions per month — 85% of which were legitimate. The false positives: costing €180,000/year in analyst time and creating merchant dissatisfaction (legitimate transactions delayed). Fraud losses: despite the rules — €1.2M in fraud losses annually. The rules: catching obvious fraud but missing sophisticated patterns that required cross-source analysis. (3) Business analytics: merchant analytics required for customer retention and growth — transaction trends, industry benchmarking, and cash flow insights. Current capability: monthly PDF reports generated manually — a data analyst spending 3 days per month compiling reports from multiple systems. The reports: 2-3 weeks old by the time merchants received them. Real-time analytics: merchants requesting real-time dashboards — competitor platforms offering this capability. The analytics gap: contributing to 4.2% monthly merchant churn. (4) GDPR compliance: transaction data containing personal information — cardholder names, account numbers, and transaction details. Current compliance: ad-hoc — personal data scattered across systems without consistent classification, access control, or deletion capability. GDPR risk: a data subject access request (DSAR) taking 3-5 days to fulfil (manual search across systems). DSARs: increasing — 15-20 per month. CNPD: increasingly active in fintech enforcement — a Lisbon fintech peer receiving a €500,000 fine for inadequate data protection. (5) Scalability: transaction volume growing 40% year-over-year — from €2B to €2.8B in 12 months. The PostgreSQL database: performance degrading as transaction volume increased — queries that took 2 seconds a year ago now taking 15 seconds. The growth trajectory: threatening to break the current infrastructure within 12-18 months.
We built a comprehensive AI data pipeline over 8 weeks using Apache Kafka, dbt, and AWS (eu-south-2). (1) Real-time transaction pipeline: streaming data infrastructure. Apache Kafka: deployed for real-time transaction streaming — every transaction event published to Kafka immediately upon processing. Event schema: transaction details, merchant information, device fingerprint, and geographic data — structured in Avro for schema evolution. Stream processing: Apache Flink processing the transaction stream in real-time. Enrichment: transactions enriched with merchant profile data, historical patterns, and risk indicators — creating a comprehensive feature set within 50ms of transaction receipt. Feature computation: real-time feature extraction — transaction velocity (transactions per merchant per hour), amount deviation (vs. merchant average), geographic anomaly (transaction location vs. merchant address), time anomaly (transaction time vs. merchant operating hours), and card behaviour (spending pattern deviation). Feature store: computed features stored in Redis (online — for real-time fraud scoring) and PostgreSQL (offline — for model training). (2) ML fraud detection pipeline: replacing rules with models. Training pipeline: historical transaction data (18 months, 45M transactions) processed through the feature engineering pipeline — creating training datasets with confirmed fraud labels. Model training: XGBoost ensemble model trained on 200+ features — transaction features, merchant features, temporal features, and card behaviour features. Model performance: AUC 0.97 (vs. 0.81 for the rule-based system), precision 94% (vs. 71% for rules), recall 89% (vs. 62% for rules). The improvement: the ML model catching more fraud while flagging fewer legitimate transactions. Serving pipeline: trained model deployed to the real-time pipeline — scoring every transaction in <10ms. Score threshold: tunable — balancing fraud catch rate against false positive rate. Model monitoring: real-time model performance tracking — precision, recall, and false positive rate monitored continuously. Feature drift: automated detection of feature distribution changes — alerting the data science team when input patterns shift. Retraining pipeline: monthly automated retraining on updated data — ensuring the model adapts to evolving fraud patterns. A/B testing: new model versions deployed alongside the production model — performance compared before full promotion. (3) Analytics data warehouse: business intelligence infrastructure. Data warehouse: dbt-managed warehouse on AWS Redshift. Source integration: data from PostgreSQL (transactions), Salesforce (merchants), Oracle (settlements), and Zendesk (support) — all integrated into a unified warehouse. Data modelling: dimensional model — fact tables (transactions, settlements, support tickets) and dimension tables (merchants, time, geography). dbt: 85 transformation models — tested, documented, and producing reliable business data. The warehouse: enabling cross-source analysis that previously required manual Excel merging. Merchant analytics: real-time merchant dashboards — transaction trends, industry benchmarking, seasonal patterns, and cash flow analysis. Dashboard update: hourly (from monthly PDF reports). The dashboards: accessible via web portal — merchants accessing their analytics 24/7. (4) GDPR compliance pipeline: privacy by design. PII classification: automated PII detection across all data sources — cardholder names, PANs (masked), account numbers, email addresses, and phone numbers. Column-level security: PII columns encrypted at rest and accessible only to authorised roles. Anonymised views: analytics and ML pipelines accessing anonymised data — cardholder PII not required for fraud detection or business analytics. DSAR automation: data subject access requests processed automatically — the pipeline querying all systems for the subject's data and compiling a DSAR response in minutes (from 3-5 days). Right to deletion: automated deletion across all pipeline stages — from raw storage through the warehouse to the feature store. Deletion audit: confirmation log verifying data removal across all systems. Consent management: consent metadata tracked — processing restricted to consented purposes. (5) Scalability: designed for 10x growth. Kafka: handling 50,000 events per second (current peak: 5,000) — 10x headroom for growth. Flink: auto-scaling processing capacity based on event volume. Redshift: scalable warehouse with automatic workload management. Cost optimisation: tiered storage (hot/warm/cold) — recent data in fast storage, historical data in cost-effective storage.
Results over 10 months. Fraud detection: fraud losses from €1.2M/year to €380,000/year (68% reduction — €820,000 annual saving). False positive rate: from 8.2% to 1.4% (83% reduction). Manual reviews: from 4,100/month to 700/month — analyst team from 6 to 2 (4 analysts redeployed to merchant risk assessment and compliance). Analyst time saved: €140,000/year. Fraud detection speed: from hours (rule review + manual investigation) to <10ms (real-time ML scoring). Sophisticated fraud: the ML model catching cross-source patterns invisible to rules — sequential card testing across merchants, velocity manipulation across time zones, and merchant collusion patterns. New fraud patterns caught: 12 novel fraud typologies identified by the model that rules had never detected. Merchant analytics: dashboard adoption by 78% of merchants within 3 months. Merchant retention: monthly churn from 4.2% to 2.8% (33% reduction). The improvement: real-time analytics giving merchants a reason to stay — competitive feature parity with larger competitors. Each 1% churn reduction: approximately €120,000 in preserved annual revenue. Total retention impact: approximately €168,000/year. Analytics-driven upsell: data-informed merchant recommendations generating 15% increase in average revenue per merchant. GDPR compliance: DSAR response time from 3-5 days to 15 minutes (automated). DSARs processed: 180 over 10 months — all within 24 hours (well within the 30-day GDPR requirement). CNPD audit readiness: comprehensive data lineage, access logs, and processing records — ready for regulatory review. Right to deletion: tested successfully — complete data removal across all pipeline stages in <1 hour. Data quality: data quality issues detected before impacting analytics or models — 47 data quality incidents caught automatically (from 0 automated detection previously). Processing performance: cross-source queries from hours (manual) to seconds (warehouse). Pipeline reliability: 99.8% uptime — pipeline processing without failure. Development cost: €120,000 (8-week project — Kafka, Flink, dbt, Redshift, GDPR compliance layer). Monthly operating cost: €4,500 (AWS infrastructure, Kafka managed service). Annual value: €820,000 (fraud reduction) + €140,000 (analyst redeployment) + €168,000 (merchant retention) + GDPR compliance risk reduction + scalability for growth. Total annual value: €1.13M. ROI: 9.4x in the first year.
Data extraction, transformation, and loading pipelines designed specifically for ML — handling feature engineering, data augmentation, and train/test splitting.
Centralized feature stores that serve consistent features to training and inference pipelines, with point-in-time correctness and real-time serving.
Annotation platforms and workflows with quality control, inter-annotator agreement tracking, and active learning to minimize labeling costs.
Ingest, parse, chunk, embed, and index documents from PDFs, Word, HTML, and other formats for RAG systems and knowledge bases.
Automated checks for data drift, schema violations, missing values, and distribution shifts that alert teams before bad data reaches models.
Real-time data pipelines using Kafka, Redis Streams, or cloud services for online feature computation and low-latency ML serving.
Lisbon's AI data pipeline market includes data consultancies (established European data consultancies — Xpand IT (Portuguese), Devoteam — offering data engineering within broader digital services. Strong local knowledge at €150-250/hour), cloud providers (AWS, Azure, GCP professional services — deep platform expertise but limited Portuguese business context and GDPR operational knowledge), freelance data engineers (Lisbon's growing freelance data engineering community — capable individuals at €60-100/hour, varying pipeline architecture experience), and internal data teams (Lisbon companies building internal data engineering — growing but struggling to recruit data engineers in Lisbon's competitive market). ZTABS provides AI data pipeline development with European compliance expertise — GDPR-by-design, EU AI Act readiness, and Portuguese data source integration — combined with production-grade pipeline engineering and ML operations.
AI data pipeline development in Lisbon ranges from €40,000-100,000 for focused pipelines (single data source, batch processing, basic ML pipeline, 4-6 weeks) to €100,000-300,000 for comprehensive data platforms (multiple sources, real-time + batch, feature store, GDPR compliance, 8-14 weeks). Monthly infrastructure: €2,000-8,000 (cloud compute, storage, and managed services). ZTABS provides fixed-price pipeline development with data quality guarantees.
Verified reviews from real client engagements — sourced from our public testimonial archive and Clutch profile.
My experience is throughout positive. Communication, service, the short response times and the flawless execution of a challenging topic was absolutely great. ZTABS is definitely my first choice again.
Christian Neff
Bank Software Advisory · Bank Software Advisory
Fintech
Fantastic Agency! I couldn't fault them even if I tried. They always go above and beyond to meet your expectations and always produces quality work. Thank you ZTABS.
Stephanie Kal
CEO · Beauty Finder Australia
Marketplace
It has been great working with ZTABS. They bounce off the ideas along the way. Amazing Experience.
Joel Rowe
CEO · Drill Quoter
Marketplace
We don't just contract — we ship and operate our own software. 17 products in production.
Common questions about AI data pipeline development for Lisbon businesses
Privacy by design — GDPR built into the pipeline architecture, not bolted on. Automated PII detection and classification, column-level security, anonymisation in the appropriate layer, right-to-deletion capability across all stages, and consent tracking throughout the pipeline. DSAR automation: data subject requests processed in minutes, not days.
ZTABS is a remote-first AI data pipeline development agency serving Houston businesses — including etl for machine learning, feature stores, data labeling workflows. We work with Energy & Oil/Gas, Healthcare & Biotech, Aerospace & Defense companies in Houston, TX via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
AI Data Pipeline Development in New York, NYZTABS is a remote-first AI data pipeline development agency serving New York businesses — including etl for machine learning, feature stores, data labeling workflows. We work with Finance & Fintech, Media & Advertising, Fashion & Retail companies in New York, NY via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
AI Data Pipeline Development in Los Angeles, CAZTABS is a remote-first AI data pipeline development agency serving Los Angeles businesses — including etl for machine learning, feature stores, data labeling workflows. We work with Entertainment & Media, E-commerce & DTC Brands, Gaming & AR/VR companies in Los Angeles, CA via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Web Development in Lisbon, PortugalZTABS is a remote-first web development agency serving Lisbon businesses — including full-stack development, progressive web apps, api development. We work with Tourism Tech, FinTech, PropTech companies in Lisbon, Portugal via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
Web Design in Lisbon, PortugalZTABS is a remote-first web design agency serving Lisbon businesses — including ui/ux design, responsive design, custom interfaces. We work with Tourism Tech, FinTech, PropTech companies in Lisbon, Portugal via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
AI Development in Lisbon, PortugalZTABS is a remote-first AI development agency serving Lisbon businesses — including llm integration & fine-tuning, ai agents & automation, rag & knowledge systems. We work with Tourism Tech, FinTech, PropTech companies in Lisbon, Portugal via timezone-aligned engineers and async workflows; we do not have a local office, and we are explicit about that with every client.
AI Data Pipeline DevelopmentLearn more about our AI data pipeline development services nationwide.
Partner with ZTABS for expert AI data pipeline development in Lisbon. Get a free consultation today.
