~/unskewdata
Building Data Systems
That Actually Work
Principal Data Engineer. Open-source contributor. Writing about software engineering, distributed systems and databases.
KafkaSparkSnowflakeBigQueryRustPythonKubernetesTerraformAirflowHudigRPCFlink
JUST LAUNCHEDMarch 2026
Introducing unskewdata Learn
Interactive learning for software and data engineering. Explore data structures, algorithms, system design, and more through hands-on visualizations built to make every concept click.
$ cat build.log
Build Log
Real systems I've designed and shipped in production. Click to expand the full breakdown.
Building a Full Data Platform from Zero, 3 Times, Across AWS and GCP
AWSGCPRedshiftBigQueryDBTAirflowTerraform
3x
Platforms Built
Cutting Time-to-Insight from Days to Minutes with Real-Time Snowflake Ingestion
SnowpipeSnowflakeAWS SQSS3StreamlitReal-Time
Days → Min
Time to Insight
Real-Time Customer Data Enrichment via Kafka Streams at Scale
Kafka StreamsKSQL DBApache HudiKubernetesRocksDB
Real-Time
Enrichment
Bridging GCP and AWS: Cross-Cloud Event Pipeline for Unified Analytics
GCPAWSPub/SubKinesis FirehoseRedshiftCross-Cloud
2 Clouds
Unified Pipeline
Zero-Downtime Event Tracking at Scale with the Outbox Pattern
FastAPIKafkaOutbox PatternProtobufHelloFresh
0
Events Lost
GDPR-Compliant Data Lake with ACID Transactions on S3
Apache HudiS3SparkGDPRAWS GlueAthena
ACID
On S3
Scaling Spark Pipelines Across Three Companies and Two Clouds
Apache SparkPySparkScalaAWS EMRGlueAirflow
3
Companies
High-Performance Vector Search API in Rust for ML Recommendations
RustAxumgRPCQdrantAWS EKSRedis
Sub-ms
Latency