Snowcap DatalabTarun Kumar — Data & AI ArchitectureWhen the architecture decisions get hard, that's when I start.
I'm an independent data & AI architect who steps in when a platform has outgrown its first design, or when there's no platform yet and you want it built right from day one — modernizing what exists, architecting what's new, making the calls your team needs, and building GenAI systems that hold up in production.
The problems that land on my desk
The platform works, but nobody trusts it anymore.
Years of tactical fixes have left it hard to extend, slow to query, or quietly wrong in places nobody's found yet.
Everyone agrees GenAI should be next. No one agrees how.
There's pressure to ship an LLM feature, but no consensus on architecture, data access, or where the guardrails go.
The team can build. The team can't decide.
Strong engineers, but no one whose job it is to own the hard calls — schema boundaries, build vs. buy, what to migrate and what to leave alone.
A rebuild sounds easier than it is.
The instinct is to start over. Usually the platform doesn't need replacing — it needs someone who's rebuilt enough of these to know which parts to keep.
Three things, done properly
Data Platform Modernization
Modernize an existing lakehouse or warehouse without throwing away what already works. I look at what's there, isolate what's actually broken, and re-architect around it — not a rip-and-replace.
Data & AI Architecture
Make the architecture decisions an engineering team needs before committing to expensive implementation — data modeling, platform boundaries, build-vs-buy, migration sequencing.
Production GenAI Systems
Design RAG, agentic, and LLM systems that fit into what you've already built — not a bolted-on chatbot. Data access, evaluation, guardrails, and the parts that make it production-grade.
Who this is for
This works well when
- You have an existing data platform that needs modernization
- You're starting a completely greenfield platform with no existing architecture or team
- Your engineering team needs senior architectural direction they don't currently have
- You're introducing GenAI into an existing product or data platform
- You want full-time architectural focus without a permanent hire
- You want someone who makes decisions and also gets hands-on
This probably isn't a fit if
- You need a large implementation team
- You're looking for an offshore development agency
Recent engagements
Unifying SAP and Salesforce contract data on a governed Databricks lakehouse
Real-time Kafka CDC to a Contract API, SOX-auditable daily runs, staged Unity Catalog migration.
→ Industrial & Tech ManufacturingIdentity synchronization and service-principal risk analytics on Databricks
First consolidated, scored inventory of the platform's service-principal population.
→ Veterinary DistributionTokenizing PII in real time on a Kafka-to-Snowflake streaming pipeline
Raw PII never in plaintext, ~60–90s end-to-end latency, 9,000+ practices served.
→ Global Manufacturing & DistributionConsolidating global SAP supply chain data on an event-driven Databricks lakehouse
Certified Gold-layer models consumed by 2,000+ daily users across regions.
→ Industrial / Electronic ComponentsIntroducing a production GenAI system on top of an existing manufacturing data platform
Hybrid vector + graph RAG platform, BYOM/BYOC, 100% data compliance maintained.
→See all case studies, including illustrative reference scenarios →
A plain engagement model
Full-time, one engagement at a time
Fractional in contract structure, full-time in practice — I take on one engagement at a time so it gets full attention, not a slice of a split week.
Me, directly — no bench, no handoffs
I don't subcontract. You work with me from the first call through implementation.
3-month minimum engagement
Architecture decisions need time to land. See the Services page for the full engagement model and pricing band.
Who's behind this
Have an architecture decision you're stuck on?
Bring me the messy version. That's usually where the real work starts.