Shaharmeer Basharat

Engineered an autonomous RAG pipeline using DB-tenant-isolated vector namespaces in Weaviate, eliminating cross-tenant data leakage and passing enterprise security audits. Built a Kafka-driven embedding pipeline to keep P99 retrieval latency under 120ms for a 1M+ active user base, and hardened security with zero-downtime migration plus PostgreSQL RLS and immutable audit logs for SOC 2 readiness.

Architected a zero-downtime ML serving layer by replacing FastAPI model wrappers with NVIDIA Triton Inference Server, enabling concurrent model execution with dynamic GPU memory isolation to prevent OOM crashes. Built a KServe canary deployment system with automated statistical A/B evaluation and deployed model drift detection integrated with Prometheus and Grafana for safer releases.

Engineered a distributed entity resolution engine to process, match, and persist millions of heterogeneous business records. Rebuilt monolithic ingestion into independently scalable microservices with Kafka and Celery, implemented exactly-once processing with idempotent consumers, and added an ML bi-encoder + pgvector (HNSW) layer to reduce duplicate rates to under 0.4%.

Built and maintained CI/CD pipelines using GitHub Actions and Jenkins to enable automated high-velocity deployment workflows. Implemented load and stress testing to validate scalability, and automated regression/functional testing integrated into the deployment pipeline for zero-downtime, fault-tolerant releases.

Developed and maintained scalable Python backend architectures using FastAPI and Django for high-availability enterprise applications. Designed end-to-end data pipelines and backend routing logic, and optimized PostgreSQL queries to minimize server response times and improve reliability.

Grade: 3.32/4.0

Earned a B.S. in Computer Science from the National University of Modern Languages (Islamabad), with a CGPA of 3.32/4.0.