Jhon Lucas · Founder & Data/AI Engineering Consultant
Data platforms, AI workflows, and blockchain infrastructure for teams that need dependable systems—not another prototype.
I like messy data, clear systems, and infrastructure that becomes boring for the right reasons.
01 / Selected work
Architecture matters, but so do recovery, deployment, observability, and the handoff to the people who operate what was built.
Case 01 · Web3 streaming
A production-style event pipeline that turns raw Uniswap activity into a restart-safe, observable stream for downstream analytics.
Handling checkpoints and chain reorganizations without corrupting downstream state.
Decoupled Kafka producers and consumers with PostgreSQL, Prometheus, and Grafana.
Python, Kafka, PostgreSQL, Docker, Prometheus, Grafana
Read the repository ↗
Case 02 · Data platform
A shared AWS data platform built to make more than 35TB of marketplace data accessible across analytics and business teams.
Bringing 30+ databases and 300+ tables into a platform that could scale without creating another operational bottleneck.
Spark on EMR with Hudi, Athena, ECS ingestion, Airbyte, Airflow, and Power BI on top of S3.
AWS, Spark, EMR, Hudi, Athena, Airflow, Airbyte, Power BI
Case 03 · Platform automation
A reusable framework that moved data work out of notebooks and into standardized, production-ready pipelines.
Cut initial setup from days to minutes and reduced pipeline delivery from roughly one month to 1–2 weeks.
YAML-driven project scaffolding and Airflow DAG generation, deployed through Docker, ECS, Terraform, and CI/CD.
Python, Airflow, YAML, Docker, AWS ECS, Terraform, CI/CD
02 / Evidence
Selected outcomes from data-platform and infrastructure work across startups, marketplaces, and Web3 teams.
years building and operating data systems
faster pipeline delivery through reusable platform patterns
cloud-cost reduction through ECS and Iceberg migrations
Spectral Labs, OpenBlock Labs, and Olist—working across platform architecture, analytics engineering, infrastructure, and technical mentorship.
03 / Work with me
I usually work through one of three formats. Each is scoped around a concrete decision, system, or operating problem.
Find the reliability, cost, and maintainability risks in an existing data platform, then turn them into a prioritized roadmap.
Usually 1 weekDesign and ship a focused production system: orchestration, transformations, an AI workflow, or event ingestion.
Usually 2–5 weeksSupport architecture decisions, production hardening, cost control, and team practices without adding a full-time role.
Part-time04 / References
“One of the strongest data engineers I've had the chance to collaborate with. His AWS knowledge is impressive, and he always figured things out fast.”
“Strong data engineering expertise, ownership, and practical problem-solving. He played a key role in building the data platform, feature store, and APIs.”
“Thoughtful, efficient, and takes ownership. He consistently delivered reliable, high-quality data infrastructure.”
“Turns complex blockchain and DeFi data into something reliable and actionable. A pragmatic partner across platform, analytics, and modeling work.”
05 / About
I enjoy turning complicated systems into dependable infrastructure that teams can understand and own.
I'm Jhon Lucas, a data and AI engineering consultant based in Brazil. Over the past eight years, I've worked across startups, marketplaces, and Web3—building platforms, automating delivery, and helping teams make better technical decisions.
I work comfortably across Python, Spark, dbt, Terraform, ECS, FastAPI, and modern AI tooling. The technology matters, but I care more about the habits that keep systems useful after the initial launch.
06 / Project archive
A compact record of platform, analytics, infrastructure, and applied-data work.
Build reliable, cost-effective transformations across Ethereum, Arbitrum, Linea, and EigenLayer without duplicating model logic.
Created reusable dbt macros and incremental models, orchestrated by Airflow across Athena, Snowflake, and BigQuery with testing and partitioning standards.
Delivered faster cross-chain insights, more reliable models, and lower transformation costs.
dbt · Airflow · Athena · Snowflake · BigQuery · Iceberg · Python · SQL
Manual exports from Dune Analytics were slow, error-prone, and prevented teams from accessing blockchain data quickly.
Built self-service ingestion into Athena with ECS, Airflow, Docker, Terraform, and Slack status notifications.
Reduced ingestion time by 90%, removed recurring manual steps, and improved reliability and onboarding.
Python · AWS ECS · Athena · S3 · Airflow · Docker · Terraform · Slack API
Move beyond a NoSQL-only setup and support blockchain, lending-protocol, and cryptocurrency data in one analytical platform.
Evaluated open table formats, designed the multi-source architecture, and implemented the platform and its analytical access patterns.
Established a functional lakehouse that made DeFi and blockchain data consistently available for analytics.
AWS · Apache Spark · Apache Hudi · Airflow · Terraform
Track unique externally owned accounts across Aave V1, Aave V2, and Compound events with changing on-chain state.
Extracted lending events, resolved unique addresses, and used Apache Hudi upserts to maintain the tracking dataset.
Created a repeatable address-tracking system for efficient DeFi event analysis.
Apache Hudi · DeFi · Python
Obtain comprehensive Ethereum transaction data for credit scoring faster and more reliably than the Etherscan API allowed.
Compared Geth and Erigon, deployed an Erigon archive node, and built ECS extraction plus Athena and EMR Serverless processing and wallet models.
Delivered a high-performance transaction data system for on-chain credit analysis.
Ethereum · Erigon · AWS ECS · EMR Serverless · Athena · Airflow
Collect lending-protocol events from the Polygon subgraph on The Graph for downstream analysis.
Configured event extraction, processed protocol data, transformed it for analysis, and connected it to the existing data platform.
Enabled detailed and repeatable analysis of Polygon lending activity.
The Graph · Polygon · DeFi · Python
An MWAA network configuration created unnecessary traffic costs and security concerns.
Reworked the VPC configuration, removed the NAT dependency, and restricted access through the private VPN network.
Reduced data-traffic costs by 90% while improving access control.
AWS MWAA · Terraform · VPC
Translate retail business rules into a dependable warehouse and reporting platform.
Worked with business and consulting teams on service deployment, dimensional modeling, and fact-and-dimension design.
Established shared business definitions and the infrastructure required for consistent reporting.
Microsoft SQL Server · SSAS · SSRS · SSIS · Targit
Frequent stockouts across retail branches were reducing product availability.
Analyzed branch inventory, encoded business rules, and built an algorithm recommending product transfers between stores.
Reduced stockouts through data-driven inventory balancing across branches.
Python · Oracle Database · SQL · Matplotlib · Jupyter
Marketing teams needed clearer customer segments for more effective campaigns.
Prepared customer data, calculated recency, frequency, and monetary scores, and combined them with cluster analysis.
Enabled targeted campaign strategies based on customer value and buying behavior.
Python · Oracle Database · SQL · Metabase
Collect and analyze Twitter activity around selected hashtags as an end-to-end academic data project.
Built ingestion with Apache NiFi and StreamSets, then processed and visualized the collected data.
Produced a working social-data pipeline and practical analysis of the captured activity.
Apache NiFi · StreamSets · Python
More than 20 AWS Lambda functions had no consistent failure-notification system.
Built Python monitoring, detailed failure messages, automated checks, and Slack delivery.
Introduced real-time operational visibility and reduced the time required to detect failures.
AWS Lambda · Python · Slack API
Make marketplace events stored in DynamoDB available for timely business analysis.
Captured changes with Kinesis Streams, processed them with Kinesis Analytics, and stored analytical data in S3 for Athena.
Delivered a real-time route from operational events to queryable lakehouse data.
AWS Kinesis · DynamoDB · S3 · Athena
Sales teams lacked a real-time path from CRM activity to analytical data.
Created a Flask webhook, processed events with AWS Lambda, deployed through Serverless, and stored the results in S3.
Made CRM events available for near-real-time sales analysis and decision-making.
AWS Lambda · Flask · Serverless · S3
Technical profile
Python · SQL · Spark · dbt · Airflow · AWS · Terraform · Docker · ECS · Athena · Iceberg · Kafka · PostgreSQL · FastAPI · AI agents · retrieval systems · blockchain data
