Architect·6 weeks·Hybrid
Apache Spark Structured Streaming in Production
Wire Kafka-aligned sources, stateful operators, and checkpoint folders without hand-waving the failure modes.
Big Data + Apache Spark · KR delivery spine
312 checkpoint recoveries rehearsed in class repos (rolling 18 months)
Hwaseong-si crews pair with our facilitators to keep Spark clusters explainable: streaming joins, Iceberg branches, and listener buses documented like production incidents—not slide fiction.
Download syllabus pack
Trusted by teams at VoltMesh Logistics, HanRiver Analytics Co-op, Paju Fabrication IoT for streaming drills.
2.6M
rows replayed in anonymized skew drills
54
cohorts since 2019 with published change logs
9.2
facilitator clarity (internal pulse, /10)
18
racks time-shared for hybrid lab windows
42
employer-authored constraints in rubrics
Quiet channel
Quarterly field notes for Spark operators
No countdown widgets—just a concise memo on watermark changes, rack firmware, and syllabus diffs whenever we cut a new intake.
Catalog surface
Programs indexed like infrastructure, not marketing arcs
Eight Spark-forward builds with filters for level, duration, study format, and certificate paths. Cards stay horizontal so you scan like a capacity planner, not a carousel shopper.
Showing 8 of 8 programs
Practitioner·5 weeks·Live studio
Delta Lake Patterns for Governed Lakes
ACID batches meet streaming merges: schema enforcement, vacuum cadence, and time travel queries you can defend in review.
Practitioner·4 weeks·Async lab
PySpark DataFrame Performance Clinic
Shuffle partitions, AQE toggles, and skew hints taught as measurable experiments—not folklore copied from forums.
Architect·7 weeks·Hybrid
Kubernetes for Spark on YARN Escape Paths
Move driver/executor placement to K8s pod templates with resource profiles tuned for Hwaseong-class bare-metal racks.
Practitioner·5 weeks·Live studio
Batch Feature Pipelines with Spark MLlib Hooks
Feature stores fed by nightly Spark jobs: point-in-time correctness, idempotent writes, and drift alarms wired to notebooks.
Architect·6 weeks·Hybrid
Apache Iceberg Catalogs on Spark 3.5
Hidden partitioning, branch snapshots, and compaction jobs orchestrated with realistic table maintenance budgets.
Foundations·3 weeks·Async lab
Metrics, Tracing, and Log Correlation for Spark Apps
OpenTelemetry exporters, Spark listener buses, and executor log stitching so on-call engineers sleep slightly better.
Foundations·4 weeks·Live studio
Probabilistic Sketches at Scale with Spark SQL
Theta sketches, HLL, and count-min sketches inside guarded UDFs with reproducible accuracy/error trade notebooks.
Execution spine
Five beats we repeat until muscle memory lands
Step 1
Frame the failure budget
Step 2
Replay with captured plans
Step 3
Instrument listener buses
Step 4
Publish diffable notebooks
Step 5
Retro with employer rubric
Signal wall
Experience-first quotes, not miracle metrics
“The Structured Streaming labs forced us to log watermark decisions we had been hiding in Slack threads. Checkpoint forensics alone justified the Mesh Foundry tier.”
Laura P. · Data platform lead · Nordic freight exchange
“Iceberg branch snapshots finally clicked after the week-two compaction drill—still slower at PySpark than I want, but honest about why.”
Ji-ho · Busan
“Mentors answered my skew hint question with a live explain plan instead of a canned deck.”
Amelia S. · verified · 4.8/5 clarity score (internal)
“Finance wanted receipts for every performance claim; the listener bus module gave us metric exports they could trace.”
Client in telecom wholesale
“Career coach helped me rewrite bullet points around the Delta Lake governance project without inventing titles I never held.”
Marco L. · Senior engineer · Regional insurer
Partnerships
VoltMesh Logistics
HanRiver Analytics Co-op
Paju Fabrication IoT
Gwangju Transit Signals Lab
Micro FAQ
Do I need a running Spark cluster on day one?
No. Week one runs against our shared lab; you only need Docker-capable hardware and VPN access we provision after enrollment confirmation.
How are refunds handled if my employer freezes training budgets?
See the Refund pathway page for eligibility windows, partial refund cases, and processing timelines under Korean consumer guidance.
Can teams from the same company share a cohort seat?
Seats are per named participant so feedback stays attributable; we can invoice multiple cost centers on request.
Artifacts
Bring the syllabus PDF into your architecture review
Includes week-by-week cluster expectations, VPN prerequisites, and the refund pathway summary so finance can sign without chasing us for footnotes.