SaaS — 8s queries → 80ms
Added a covering index, replaced a N+1 with a single join, materialized a 3-table aggregation. p99 dropped from 8s to 80ms. No code change required on the application side.
The database layer your data team can scale on.
Postgres, ClickHouse, time-series, OLAP. Schema design, query tuning, replication, backup, and the boring operational discipline that makes a database survive its first 10× growth.
Rate $90 – $120 USD / hour · fixed-fee and retainer available
What you get
- Schema design
- entity-relationship modelling, normalization vs denormalization, indexing strategy, constraint design.
- Query optimization
- EXPLAIN ANALYZE, hot-path tuning, index health, materialized views where they actually help.
- Replication & HA
- streaming replication, logical replication, Patroni, pg_auto_failover, multi-region.
- Backup & PITR
- pgBackRest, WAL-G, point-in-time recovery, tested restores (not just taken backups).
- Migration
- online schema changes (pgroll, pg_repack), zero-downtime cutovers, rollback plans.
- Observability
- pg_stat_statements, pgAudit, slow-query log, connection pooler tuning.
How we engage
- 01
Performance audit (1 week)
We trace your slow queries, look at your schema, and write a prioritized report. Most startups find 3–5 quick wins that recover 80% of lost performance.
- 02
Reference build (2–4 weeks)
One end-to-end migration or a greenfield schema that becomes your team’s template.
- 03
On-call DBA retainer
We cover your database on-call — incidents, schema reviews, capacity planning.
Stack we work in
OLTP
PostgreSQL, MySQL, MariaDB, Aurora, Cloud SQL, AlloyDB
OLAP
ClickHouse, BigQuery, Snowflake, Redshift, DuckDB
Time-series
TimescaleDB, InfluxDB, QuestDB, Prometheus, M3DB
Cache
Redis, Valkey, KeyDB, Memcached
Search
OpenSearch, Elasticsearch, Meilisearch, Typesense
Migration
pgroll, pg_repack, gh-ost, pt-online-schema-change
Reference architectures
Anonymized patterns from real engagements. Client names omitted; details available under NDA.
Logistics — 50M rows/day, 6-month retention
Migrated from Postgres to ClickHouse for the analytics path. Postgres kept for OLTP, ClickHouse for OLAP. Storage cost down 60%, query latency down 95%.
Healthtech — 4-region replication, RPO < 1 minute
Streaming logical replication across 4 regions with automated failover. Tested quarterly. Cleared HIPAA on the first audit.
Questions we get asked
Postgres or MySQL?
Postgres for almost everything. MySQL only when you’re inheriting a MySQL codebase, or when you have a very specific operational reason.
Do you help with ClickHouse?
Yes — ClickHouse is one of our core specialties. Schema design, materialized views, sampling, retention.
Can you take over a struggling database?
Yes. We start with a 1-week audit, find the immediate risk, and ship a stabilization plan in the first month.
Do you do on-call?
Optional — 24/7 database on-call with a 15-minute response SLA. Most clients use us during US business hours and their own team covers the rest.
What’s your take on ORMs?
Use them for simple CRUD. Hand-write SQL for hot paths. We review the data access layer for the 5–10 queries that actually matter.
Let’s scope it properly.
A 30-minute call. No deck, no pitch — we read your repo or your architecture diagram and tell you what’s realistic.