Backend Engineer specializing in distributed systems and AI-native platforms at Oracle. I build multi-tenant, cloud-native services on OCI — hybrid RAG search with ACL-filtered, citation-backed answers, cache-backed JDBC repositories (40% queries migrated, P99 −60%), and cross-region observability (OpenTelemetry/OCI APM). I ship high-performance backends in Java/Go/Python with Kafka, Oracle Coherence, and K8s, and have modernized the Java Client SDK from Pekko to a custom async HttpKit. Also building AI products end-to-end — Next.js CV anonymization tool and multi-cloud object storage facades.
Multi-tenant distributed facade for object storage (OCI Buckets, AWS S3) on OCI — hybrid RAG search, cache-backed data layer, and zero-downtime global platform upgrades.
Designed secure RAG search architecture for a multi-tenant file platform using hybrid retrieval (vector embeddings) with ACL-filtered, citation-backed responses.
Refactored DB access into JDBC and cache-backed repositories (Oracle Coherence, Autonomous DB); migrated ~40% of queries to repository pattern, cutting P99 latency by ~60% vs baseline.
Diagnosed and resolved distributed-cache deployment failures across REST cache access, Istio, Helm charts, and node-group configuration.
Replaced deprecated auth layer with OCI API Gateway — OIDC routing and deployment automation for cloud-native environments.
Built Kubernetes-based DB observability with a custom exporter CRD and OCI Metrics Explorer validation.
Implemented cross-region distributed tracing (OpenTelemetry, OCI APM, OCI policies) to distinguish multiple service instances within the same region.
Modernized Java Client SDK by migrating from Pekko to internal HttpKit, preserving async APIs, OAuth, timeouts, and error handling.
Next.js tool that anonymizes, reformats and enhances CVs into consistent, editable PDFs — local-first agentic extraction with multi-step LLM refinement.
Parses PDF/DOCX, strips PII (emails, phones, links) while preserving signal — first-name-only retention when needed.
Reformats to a single clean layout with rich-text editor for human-in-the-loop edits and PDF export.
Multi-step AI workflow for enhancement + language refinement using OpenAI structured outputs, LangChain/LangGraph.
