Software engineer · Hyderabad, India

B Prithiraj

Building dependable systems,
from backend services to AI workflows.

I work on Java services, distributed data, and AI-assisted workflows at Oracle. Outside work, I build inference infrastructure and explore how software should behave when requests fail, state changes, and decisions need an explanation.

Read my resume GitHub Get in touch

Selected work

All projects

InferenceMesh
Working implementation
A reproducible inference gateway that bounds waiting work, explains route selection, and streams OpenAI-compatible responses through a deterministic backend.
Python · FastAPI · Model Serving · Observability
Read case study
PavePath
Working implementation
A Cloudflare change-readiness copilot with deterministic policy decisions, AI explanations, durable review history, and an explicit human exception workflow.
TypeScript · Cloudflare · Durable Objects · Workers AI
Read case study
Aegis ResponseNet
Interactive prototype
An interactive reference model for command idempotency, outbox replay, read-model convergence, and compensation after a downstream failure.
TypeScript · Idempotency · Event-Driven Systems · Sagas
Read case study Try demo
FinAgent RiskOps
Interactive prototype
An interactive agent-workflow reference model with synthetic evidence, explicit tool permissions, cited answers, and deterministic review or stop decisions.
TypeScript · Agent Workflows · Evaluation · Tool Permissions
Read case study Try demo

Engineering notes

All writing

Failure cases, design decisions, and the trade-offs behind reliable software.

An inference gateway needs a queue limit and a stream owner
Why concurrency limits alone do not protect an inference API, and how response ownership changes once a token stream starts.
7 min read · Model Serving · Python
Keep policy decisions separate from LLM explanations
A practical boundary for AI-assisted workflows: versioned facts and deterministic decisions, with an explanation that cannot grant permission.
6 min read · AI Systems · Policy
The stale cache write that happens after invalidation
Deleting a cache entry does not stop an older reader from putting it back. Monotonic versions make that race explicit and testable.
6 min read · Caching · Distributed Systems
A timeout is an unknown outcome, not a failed operation
How idempotency, reconciliation, and version checks make recovery safer when a remote operation may have committed before the caller timed out.
6 min read · Distributed Systems · Reliability

Let’s talk systems.

I’m interested in backend engineering, distributed systems, and AI infrastructure. If that’s what you’re building, I’d like to hear about it.