Abhyush Rajak

About

Backend/event-driven systems engineer who built an end-to-end “software robot” for AI-driven invoice processing: FastAPI ingestion + OCR integration + classification mapping, with strong emphasis on reliability (idempotency, retries) and scalability (background workers, event-driven architecture). Experienced in production-grade distributed systems tooling (Kafka, Docker/Kubernetes, GitHub Actions, ArgoCD) and real-time debugging via tracing/telemetry, and expects $10k–$12k/month.

Experience

Education

Key Strengths

• Designed and owned backend APIs for an AI/OCR-driven invoice processing pipeline
• Improved data quality by adding preprocessing and validation layers for inconsistent OCR outputs
• Built reliable workflows with retries and idempotent processing to prevent duplicate work
• Scaled high-concurrency ingestion by offloading heavy OCR to background workers and using event-driven processing
• Debugged tight-latency, real-time event pipelines using trace IDs, structured logs, and telemetry to diagnose ordering/timing issues
• Designed cross-service communication via shared Kafka message schemas with versioning, DLQs, and resilient consumers
• Owned end-to-end backend build for invoice ingestion and automated tax calculation (FastAPI + SQLAlchemy)
• Improved OCR accuracy via image preprocessing and post-extraction validation (totals/required fields/keyword mapping)
• Production deployment experience with Docker and AWS (EC2, S3)
• Kubernetes deployment with autoscaling plus GitOps CI/CD using GitHub Actions + ArgoCD (health checks, rollbacks, drift detection)
• Planned and executed on-prem to AWS migration using Terraform with staging dry runs, backups, load testing, and DNS cutover
• Built Kafka-based real-time event pipeline with schema usage, retries, idempotency, buffering for out-of-order events, and DLQ handling