Open to conversations on architecture, agentic AI & the AI-native SDLC
Chaitanya Sunkara — Software architect & consultant

Architecture that survives the next framework, and the next one after that.

I help teams design systems that hold up in production — cloud & microservices architecture, agentic AI and AI-led automation, and the AI-native SDLC. 19+ years across .NET, SQL and the cloud, applied to the questions that actually matter: what's the real problem, where does complexity belong, and what breaks at scale.

Subscribe for updates → Read recent writing · or scroll for the short version
Practice
Architecture
+ agentic AI & automation
Experience
19+ years
enterprise .NET & SQL
Writing
Since 2016
notes from the field
Currently
Writing & building
happy to trade notes
/what I do

Four practices, one underlying discipline.

Each of these is an area I've been paid to lead — not a list of technologies I've touched. The through-line is logic-first thinking: surface tech changes, the reasoning doesn't.

01 — Practice

Enterprise & cloud architecture

Reviews, designs and second opinions for teams betting meaningful money on a technology choice. I look at where complexity actually belongs, what will break at scale and at handover, and which of your "must-haves" are really constraints inherited from an earlier decade. Strong fit for platform rewrites, acquisition integrations and cloud transitions.

02 — Practice

Agentic AI & AI-led automation

Designing agentic systems that do real work inside real enterprises — not demos. The hard parts are rarely the model: they're tool boundaries, state, guardrails, retries and the handover between agent and human. I help teams ship automation that's trusted by operators and understood by auditors, built on top of the cloud architecture you already have.

03 — Practice

AI-native SDLC

Embedding AI into the software development lifecycle without breaking the lifecycle. Code generation, review, testing and operational knowledge graphs — evaluated for where they genuinely compound engineering leverage and where they introduce silent debt. I help technical leaders separate the signal from the keynote noise.

04 — Practice

Secure microservices & modernisation

19+ years of pattern-matching on the specific way enterprise .NET monoliths, SQL Server estates and legacy analytics stacks turn into event-driven, cloud-native platforms. I've led .NET Framework → .NET Core, SQL Server → Aurora / Snowflake, on-prem → AWS migrations across multiple clients. Security and production-readiness audits are part of the same practice.

/selected work

Recent engagements, in outline.

Architecture reviews, agentic-AI design, data modernisation, secure microservices. Year to year, the surface varies; the underlying discipline doesn't.

Enterprise SaaS

Agentic AI platform for back-office operations

Architected a multi-agent platform that plugs into an existing microservices estate — specialised agents, orchestration, audit trail, event-driven tool layer. Controller → Service → Repository boundaries extended cleanly for AI.

Production pilot with auditor sign-off and human-in-the-loop handover by design.
agentsorchestrationeventsaudit
Enterprise SaaS

Natural-language query layer over business data

Designed the intent / entity / retrieval pipeline for a natural-language interface sitting on top of an existing business-services stack. Multi-provider AI, conversation state, feedback loop, cost telemetry — all within established microservice boundaries.

Unified query surface for non-technical users without forking the platform.
RAGintentmulti-provider
Enterprise platform

AI-native SDLC pilot for a product engineering org

Designed the evaluation framework, tool selection, guardrails and rollout plan for embedding AI across code generation, review and test automation — without compromising the existing release process or security posture.

Clear compound-leverage areas identified; signed off by security.
SDLCevalssecurity
Analytics / data

Legacy SQL Server estate → Aurora & Snowflake

Led the migration of a mixed reporting and OLTP estate onto Aurora Postgres and Snowflake. Rewrote the SSRS / PowerBI layer against the new warehouse without a reporting blackout, using slice-by-slice cutover.

Slimmer infrastructure footprint; report latency materially reduced.
AuroraSnowflakePowerBI
/writing

Notes from the field, thinking in public.

Architecture decisions explained in plain prose. Written mostly for engineers and technical leaders deciding where to spend complexity.

All writing →
/parikrama · परिक्रमा

A pilgrim's atlas of Hindu temples.

Two-thousand-plus temples, curated against open data, on a quiet static-first interactive map. Filter by deity, sampradaya, pilgrim circuit or architectural style; save your own yatra. A personal project, and a working sandbox for the same patterns I bring to client work.

Open the atlas

Small tools, given away

Free, browser-first utilities I built for myself and kept public — JWT decoder, LLM token estimator, cron explainer, SQL formatter, JSON formatter, and a SQL query helper. No ads, no accounts, nothing leaves your browser.

jwt-decodertoken-estimatorcron-explainersql-formatterjson-formattersql-helper
Open the tools →

Let's trade notes.

I'm always up for good conversations about architecture, agentic AI and the AI-native SDLC. Knowledge exchange, a reference, a second opinion, or just a thoughtful connection — all welcome. If you write, I reply within a day or two.