// about
Background
Independent engineer focused on backend systems, applied AI infrastructure, and the operational concerns that determine whether software actually runs: observability, reliability, and cost.
Computer engineering at APSIT, Mumbai University (CGPA 8.3). Recognized in AI for Bharat top 30. Previously collaborated on WalkSense. Currently spending time with systems programming in C — concurrency primitives, process isolation, and event-driven I/O.
Work product is shipped end-to-end: architecture, implementation, deployment, and the dashboards and runbooks that keep it operable.
// technical surface
- Backend
- FastAPI · Express / TS · Hono · REST · gRPC
- Applied AI
- Multi-agent (ADK) · RAG · Cross-encoder rerank · Embeddings · Pinecone
- Infra & DevOps
- AWS (EC2, S3) · Cloudflare Workers · Docker · Nginx · GitHub Actions
- Databases
- PostgreSQL · Redis · SQLite · D1
- Observability
- Prometheus · Grafana · Tempo · OpenTelemetry · Structured logging
- Security
- OAuth2 PKCE · JWT / Web Crypto · RBAC · tmpfs secret isolation
// principles
- 01
Operability is a feature. A system isn't done until it can be observed, debugged, and recovered by someone who didn't write it.
- 02
Write the decision down. ADRs and design docs make trade-offs auditable months later.
- 03
State of the world lives in one place. Caches and brokers accelerate; they don't own truth.
- 04
Failure modes are designed in, not patched in after the first incident.