PostgreSQL & Data Infrastructure Consulting

I help engineering teams design, scale, and optimize their databases and data platforms. My consulting focuses on two distinct areas:

• PostgreSQL Operations & Scalability: Query tuning, schema design, connection pooling, and isolating production performance bottlenecks.
• Data Pipelines & Infrastructure: Scaling query engines (Trino, Spark), designing Change Data Capture (CDC) replication, and optimizing data lake storage to reduce compute and cloud costs.

Whether you are serving data to AI agents, building real-time analytics, or modernizing your data platform, I help improve system performance, reduce infrastructure costs, and establish scalable architectures.

I have built and scaled production data platforms at LinkedIn, YugabyteDB, and StarTree. My work spans architecture reviews, performance tuning, production readiness, and scaling distributed data systems.

Technical Background

I have spent my career building, scaling, and contributing to database engines and data platforms:

• YugabyteDB: Implemented transactional streaming CDC, OpenTelemetry query tracing, and Active Session History.
• StarTree: Worked on real-time analytics query performance and scaling Apache Pinot for high-throughput, low-latency workloads.
• LinkedIn: Led big data infrastructure observability initiatives and consulted on database scaling and ETL optimization.
• Qubole: Led the SQL teams for Hive and Presto services, founding the data engineering team and scaling usage.

Who Hires Me & Why

I partner with teams facing database scaling or data architecture challenges:

• AI-Native Startups Building Production Systems: Startups needing to serve contextual data to AI agents with sub-second latency, prevent connection exhaustion from stateless agent workloads, or integrate search indexes without introducing database complexity.
• SaaS Companies Scaling PostgreSQL: Teams experiencing connection saturation, vacuum bottlenecks, write amplification, or latency spikes on PostgreSQL during peak traffic.
• Engineering Teams Migrating or Modernizing: Organizations moving away from legacy BI bottlenecks, migrating databases to the cloud, or transitioning from expensive cloud data warehouses to cost-effective open data lakes (Trino, Apache Iceberg, Spark).
• Teams Building Ingestion & CDC Pipelines: Teams setting up real-time ingestion, synchronizing search engines with operational PostgreSQL tables using Change Data Capture (CDC), or structuring processing pipelines for unstructured data.

Key Outcomes: How I Help Your Business

I focus on concrete engineering outcomes that directly impact your performance, reliability, and cloud spend.

1. Serving Data to AI Agents with Low Latency

AI agents require real-time, low-latency access to database context. I help you:

• Optimize schema layouts and index strategies for fast retrieval.
• Implement connection pooling (PgBouncer/Odyssey) and query routing to handle stateless, highly concurrent agent workloads.
• Avoid database lock contention and query latency spikes during agent execution.

2. Scaling PostgreSQL & Analytics Database Infrastructure

Complex analytical and LLM workloads can degrade your core application database performance. I help you:

• Isolate analytical workloads from transactional traffic using read-replicas and logical replication.
• Diagnose and resolve query bottlenecks using active session monitoring and database execution tracing.
• Scale PostgreSQL write throughput, configure vacuum parameters, and structure partition layouts for large datasets.

3. Architecting Reliable Data Pipelines for LLM Applications

Unstructured data ingestion and search index synchronization must be robust and low-latency. I help you:

• Design reliable Change Data Capture (CDC) pipelines using Debezium or logical replication to sync your database state to downstream search indexes.
• Streamline ETL/ELT pipelines to process ingestion batches faster and reduce pipeline lag.
• Set up observability and monitoring across data pipelines to catch schema drift and ingestion failures before they affect downstream applications.

4. Reducing Infrastructure & Compute Costs

Open-ended querying and massive analytics can cause cloud costs to skyrocket. I help you:

• Optimize query execution profiles in Apache Spark and Trino to cut down on compute requirements.
• Design storage layouts, partitioning schemes, and catalog metadata on Apache Iceberg data lakes to reduce storage and query costs by up to 50%.
• Implement caching strategies and query reuse patterns to avoid re-evaluating expensive datasets.

How We Can Work Together

I prefer focused, high-impact engagements rather than open-ended arrangements:

1. Operations & Performance Audits: A 1-2 week review of your database metrics, schema designs, or ETL execution profiles to provide an optimization roadmap.
2. Infrastructure Design & Setup: Partnering with your team to design, bootstrap, and deploy new Postgres setups, CDC pipelines, or data lake architectures (Iceberg/Trino).
3. Technical Advisory: Collaborating with your team as a fractional expert on data architecture, scalability reviews, and query engine configuration.

Contact

If you have a problem you would like to discuss, email me at [email protected] or find me on LinkedIn.