We start from measurements
We ask for AWR, execution plans, wait events and application logs first. The remediation plan gets written after the bottleneck has been pinned to a specific place in numbers.
We take on database operations and AI data platforms, under contract.
Managed DB operations and high-availability setups, performance diagnosis, heterogeneous migration, data architecture, and RAG/LLM pipeline design. We apply the same standards we used running mission-critical databases.
- Operations
- Redundancy, backup and recovery, and zero-downtime change across commercial and open-source RDBMS and NoSQL
- Domains
- Finance and securities, internet banking, retail, manufacturing, IoT, global services
- Cloud
- Data platforms on AWS RDS and Glue (Spark), and GCP BigQuery
How we work
We only propose what we have run
Redundancy and failover, zero-downtime migration, and the locking, retry and publish gates inside a RAG pipeline are proposed only in the shape we have actually run, in production or in our own systems. Configurations we only know from documentation stay off the list.
Security terms are settled before kickoff
NDA, read-only account separation, access paths and the scope of logs we keep are agreed before the contract. We build the process so that data handed over during diagnosis does not stay on our side.
Scope is fixed in writing
Deliverables, duration and the way we work are fixed up front, and we work inside that. Pick whichever fits: monthly managed operations, or project-scoped consulting.
What we cover
DBMS operations · high availability
Redundancy, failover, backup/recovery and online changes operated in production across commercial and open-source RDBMS and NoSQL.
- Redundancy · failover
- Clustering · replication · sharding
- Backup · recovery rehearsal
- Online changes
- Monitoring · capacity
- NoSQL · cache
Performance · migration
Tuning from execution plans, indexes and wait events; heterogeneous migrations with integrity validation.
- SQL tuning · index redesign
- Heterogeneous DBMS conversion
- DW → analytics platform
- Cloud migration
- Version upgrades
- Automated integrity validation
Data architecture · platforms
From modeling to standards, metadata and governance, through DW/DM and cloud big-data environments.
- Logical · physical modeling (ER-win)
- Data standards · metadata (DA#)
- DW / DM · ETL
- AWS RDS · Glue
- GCP BigQuery
- DQC · governance
AI · automation
GenAI service data structures, RAG pipelines and operations automation built both in client work and in our own research.
- GenAI metadata DB design
- RAG ETL · OpenSearch
- Local LLM pipelines (MLX)
- Python · shell automation
- Spark · Pandas
- Telegram · Slack bots
Services
Each service has defined deliverables. "Select" fills the request form below with that service. Starting prices exclude VAT; the final amount is set in the proposal after scoping.
Managed DBA Operations
Monthly operations: incident response, redundancy and failover checks, backup and recovery, capacity, privileges and release support. Any commercial or open-source DBMS, to mission-critical operating standards.
DeliverablesMonthly operations report · check lists · incident records
Starting atFrom KRW 500K / month
Performance Diagnosis & SQL Tuning
Analysis of AWR, execution plans, indexes, wait events and query patterns to pinpoint bottlenecks and deliver a prioritized improvement plan.
DeliverablesDiagnosis report · Top-SQL tuning list · before/after comparison
Starting atFrom KRW 5M
Migration Consulting
Planning and rehearsing version upgrades, cloud migration, heterogeneous DBMS conversion and HA transitions. Automated integrity validation shortens migration time and reduces errors.
DeliverablesMigration plan · risk register · rehearsal results · rollback procedure
Starting atFrom KRW 20M
Data Architecture & Standards Consulting
Logical/physical modeling, data standard glossaries and code schemes, metadata management and DW/DM design — grounded in manufacturing MES/ERP, retail DW and global-system standardization work.
DeliverablesERD · standard glossary · metadata framework · data-flow diagrams
Starting atFrom KRW 20M
AI Workflow Automation Consulting
Selecting AI candidates among repetitive tasks, document processing, internal knowledge search and support assistance, estimating impact, and designing a PoC.
DeliverablesCandidate assessment · impact estimate · PoC design
Starting atFrom KRW 10M
RAG / LLM System Design
Retrieval-augmented generation on internal data with privilege separation, cited evidence, log evaluation and operational monitoring — applying our GenAI metadata database and structured/unstructured RAG ETL reference designs, plus local-model operations know-how.
DeliverablesArchitecture design · metadata DB model · data pipeline definition · evaluation criteria
Starting atFrom KRW 20M
Technical Advisory Contract
Architecture reviews, incident-recurrence prevention and adoption decisions for CTOs, data teams and operations teams — on a recurring or per-case basis.
DeliverablesReview memo · decision notes · recurrence-prevention plan
Starting atFrom KRW 1M / month
Work Samples
1. LLM-Based Relationship Inference ERD
When production databases lack complete physical foreign keys, we review table names, column names, key patterns, and relationship candidates to build an ERD draft. The sample below shows only the inferred relationship flow with column details hidden.
Schema-first draft
Tables and key patterns are placed first so the overall structure is visible quickly.
Inferred FK flow
Physical FKs and inferred FK candidates are separated for reviewable evidence.
Partial preview
Only part of the deliverable is exposed to show the shape of the final work product.
2. Estimated effect of AI adoption
The figures below are assumptions, not measurements. Frequencies and durations are set against a typical operations team, and we recalculate them against the client's own statistics during the assessment. The scope is not total process time — only the part where a person still has to make a judgement after scripted automation has done its work.
Around 10–20h / mo
Human decision time saved (estimate on assumptions)
Roughly halved
Time spent on a single judgement
6 areas
SQL, root-cause analysis, support, tickets, reports, logs
Before is the time a person still spends after scripted automation; After is the time an LLM takes for a first pass plus human validation. Both are written as ranges.
| Work Area · Human Decision Node | Frequency | Before | After |
|---|---|---|---|
| Slow Query / Top SQL Candidate Analysis Execution-plan interpretation, rewrite options, and index alternatives | 8 / mo | 60–80 min | 25–35 min |
| Incident or Performance Event RCA Candidates Linking logs, alerts, and change history into likely causes | 4 / mo | 45–60 min | 20–25 min |
| Internal Support Based on Manuals Intent recognition and synthesis across multiple documents | 30 / mo | 12–15 min | 4–6 min |
| Unstructured Operations Ticket Classification Type, severity, and duplicate assessment from free-form requests | 40 / mo | 10–12 min | 3–5 min |
| Monthly Operations Report Drafting Interpreting metric meaning and drafting narrative actions | 1 / mo | 150–180 min | 55–70 min |
| Log Anomaly Pattern Interpretation Assessing pattern meaning and whether action is required | 12 / mo | 20–30 min | 8–12 min |
Relative comparison against each row's Before value; bar lengths use the midpoint of the range. Before and After both count human judgement time only.
Beyond Time Savings — Quality and Coverage
- SQL and RCA Junior staff can produce senior-level first-pass judgment, reducing dependency on a single expert.
- Internal Support More consistent answers with current manuals reflected automatically.
- Ticket Classification More stable classification rules and fewer missed related incidents.
- Log Interpretation Hypotheses are proposed even for patterns people may overlook, expanding analysis coverage.
How an engagement runs
1
Request
Send the service, system environment, problem and preferred timeline through the form below; we review and get back to you.
2
Scoping
We define diagnosis scope, deliverables, duration, security conditions and delivery mode (remote or on-site) in a proposal.
3
Contract & kickoff
After NDA and contract, work starts with separated access privileges. Progress is reported at the agreed cadence.
4
Delivery & handover
Reports, configuration and code are delivered with procedures and checklists so your operations team can take over.
Frequently asked questions
What contract forms and prices are available?
Managed DBA operations run on monthly contracts from KRW 500K / month. Diagnosis, migration and design are scoped project contracts: performance diagnosis & SQL tuning from KRW 5M, AI workflow automation consulting from KRW 10M, and migration, data architecture & standards, and RAG / LLM system design from KRW 20M each. Technical advisory is available on a recurring or per-case basis, from KRW 1M / month. Starting prices exclude VAT; the final amount is set in the proposal after scoping.
How are security and data access handled?
An NDA is signed before kickoff and only the minimum privileges needed are granted separately. Diagnostic data is handled within the agreed scope and, as a rule, does not remain on our systems.
Can the work be done remotely?
Most diagnosis and design work is done remotely where remote access is possible. Restricted environments are handled on-site, arranged during scoping.
Can the effect of AI adoption be estimated in advance?
As in the table above, we isolate the time a person spends judging and estimate it as a range, then measure the real values in a PoC and recalculate. Every number on this site is an assumption.
Service Request
We review each request and get back to you. A short description of the current system, the problem and the goal speeds up scoping.