Solution Architect (Hands-on) - AI-Native Bookkeeping Platform

About the role

We are building an AI-native bookkeeping and compliance platform for Indian Chartered Accountants and MSMEs.

You will own the technical design of the platform. This covers how data flows and how integrations behave. It also covers how AI components fit in. It also covers how we keep financial data accurate and auditable.

You will also build it. Expect to spend about half your time writing production code, with the rest on design, reviews and guiding the team. Not a slide-deck architect.

Our stack: Java 25, Spring Boot 4, PostgreSQL 18, Kubernetes on Azure, Terraform, Keycloak, React 19, TypeScript, Nx monorepo, OpenTelemetry. We run current versions and upgrade deliberately.

Role & responsibilities

Development

1. Build core modules yourself: ledger, multi-tenancy, integrations, AI components.
2. Write production code in Java and Spring Boot; contribute to TypeScript and React where needed.
3. Own code quality on your work: tests, CI, observability, documentation.
4. Debug and fix production issues on the parts you own.

Architecture and design

1. Own end-to-end solution design: data model, service boundaries, APIs, integration patterns, deployment.
2. Design the ledger and financial data layer so every number is traceable, auditable and reconcilable.
3. Design multi-tenancy and the control plane: tenant isolation, per-organisation routing, data separation.
4. Define how AI fits in. Decide where LLMs are used and what stays rule-based. Decide how AI output is checked before it touches the books. Decide how the MCP agent surface is exposed safely.
5. Design integrations with GST systems, banks, Account Aggregator and Tally. Handle sync, duplicates, failures and idempotency.
6. Decide how background work scales (bank feeds, OCR, reconciliation) and when it moves out of the main service.
7. Set standards for security, access control, data privacy and DPDPA readiness.

Working with the team

• Review code and run design reviews. Maintain architecture decision records.
• Turn product requirements into technical design, with the BA and product team.
• Break designs into implementable pieces for sprint planning.
• Flag technical risk, dependencies and effort early.
• Mentor developers through pairing and reviews.

Preferred candidate profile

What we need

1. 8+ years in software engineering, with 3+ years owning system design end to end.
2. Active hands-on coder: you write production code regularly today, not just review it. Recent work you can walk us through.
3. Deep Java and Spring Boot: built and run production SaaS, not internal tools. Comfortable with Spring Security, OAuth2, and staying on current versions.
4. Multi-tenant SaaS and data depth: tenant isolation, control plane and PostgreSQL data modelling. Experience with transactional or financial systems. In these systems, correctness and audit trail matter.
5. Kubernetes and cloud infrastructure: Helm, Terraform, CI/CD, observability. Any major cloud; Azure is a plus, not a filter.
6. Practical AI engineering: LLM integration, RAG, evaluation, cost and latency control. You know where AI works and where it does not.

Good to have

1. AI agent systems: Spring AI, MCP, tool use, human-in-the-loop design.
2. Identity platforms: Keycloak, OIDC, SSO, role and permission modelling.
3. Indian fintech context: GST APIs, Account Aggregator, e-invoicing, banking integrations.
4. Modern TypeScript and React: enough to build and review frontend features.
5. Security and compliance: DPDPA, ISO 27001, SOC 2, data residency.

Not required

• Deep accounting knowledge. Our domain expert and BA cover that.
• Enterprise architecture certifications.
• Experience with the exact versions we run. We expect you to pick them up.

Work setup

Fully remote, India.