AI is on the roadmap. The data layer isn't ready for it.
Inteledyne designs and builds the data platforms AI runs on. A principal architect leads every engagement, and delivery engineers build it with your team, on AWS, Snowflake and the major clouds.
Selected client experience- University of Wisconsin
- Salesforce
- Viasat
- Taco Bell
- Blue Shield of California
- Bank of America
- Hashmap / NTT Data
- 3M
accuracy classifying PII across a data lake with Bedrock agents
fewer API calls for a Bedrock agent that analyses CloudWatch metrics
to onboard a new data source, down from weeks, with metadata-driven onboarding
orchestration servers to run after replacing Airflow with events and state machines
It usually starts with a problem you can already name.
Five situations bring most clients to us. Each one maps to an engagement that fits it. See all fifteen
Most architecture advice never gets built.
We run engagements so the design gets built, and so your team can run it after we leave.
Principal-led architecture
One named principal owns the design from the first assessment to the handover, and writes down every decision.
Engineers who build it with you
Delivery engineers work in your repositories and ceremonies. Build capacity scales through our development partner, Reenbit.
AI built on governed data
Agents and retrieval go into production on a data layer that is clean, current and access-controlled.
Five practices, from the cloud foundation to the agents on top.
Each practice is led by a principal and staffed to the engagement. Most platforms need two or three of them at once.
A large commitment should start with evidence.
A first conversation, free
Bring the problem. If we are the wrong firm for it, we say so and name someone better placed.
A written answer
A Health Assessment or an Architecture Review: fixed scope, delivered as a document your board can read.
Renew, close or hand off
Embedded leadership and delivery teams are reviewed every quarter. You decide whether the work continues.
Most AI pilots stall at the data layer.
A retrieval demo needs clean, current, governed data before it becomes a production system. We build both halves: the agents, and the platform they read from.
PII classification agents
Bedrock agents classify sensitive data across the data lake at 95%+ accuracy.
SalesforcePipeline-failure triage
An LLM system finds the root cause of a failed pipeline and recommends the fix.
SalesforcePlain-English queries
Claude on Bedrock writes Athena queries from a question, runs them, and explains the answer to executives.
Axium · in-houseMetrics analysis agent
A Bedrock agent analyses CloudWatch metrics with 98% fewer API calls.
University of WisconsinData lakes, AI agents and migrations, already in production.
Designed and built a consolidated enterprise data lake on AWS EMR and Apache Iceberg for a major research university — turning complex on-premises Workday and PeopleSoft data into queryable, relational tables through an auto-relationalization engine.
Architected the strategic AWS data-platform-as-a-service originating in Marketing and progressively adopted by roughly ten additional internal teams on the trajectory toward enterprise-wide use.
Designed and implemented sub-second microservice architecture, plus the GCP migration target architecture and AI-assisted CI/CD model.
You know who is accountable before the work starts.
Robin Tanner Industries where the data is regulated, high-volume or hard to change.
- Higher educationUniversity of Wisconsin
- Enterprise softwareSalesforce
- Aerospace & satellite communicationsViasat
- Retail & restaurantsTaco Bell
- HealthcareBlue Shield of California, Healthcare Partners of Nevada
- Financial servicesBank of America, SEI Investments
- Networking & hardwarePoly/Plantronics, Juniper Networks
Running a lakehouse costs more than it should.
- How S3, Lambda, EMR and Iceberg assemble into a production lake
- The four things you have to get right
- Where Bedrock and agents plug in on top
Get the next brief by email.
A short note when a new field brief publishes: the point, and the link. One-click unsubscribe.
You get one confirmation email first. Nothing sends until you click it.
Lake AI, the product that grew out of the practice. It packages the patterns from our engagements: data foundations that absorb source-system change instead of breaking on it.
Visit Lake AI ↗Bring us the problem you can already name.
The first conversation is free. If we are the right firm for it, we propose a scope in writing. If we are not, we tell you who to call.