AI Systems Architect · M.Sc. AI Researcher @FAU Erlangen-Nürnberg
Technical Operations @Astrameld (International / Pakistan) · Erlangen, Germany
I design high-throughput AI systems, retrieval pipelines, and autonomous agent architectures that bridge cutting-edge research with production-grade engineering. My work focuses on dense vector embeddings, SIMD acceleration, multi-agent sandboxing, and evaluation frameworks for LLM retrieval systems.
Currently conducting research as an M.Sc. in Artificial Intelligence student at Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) while directing technical operations for Astrameld (serving enterprise and mid-market partners globally across US/UK/GCC).
- fast-cosine-rs — High-throughput, zero-dependency cosine similarity and vector normalization in Rust with SIMD auto-vectorization (>1.1M vectors/sec).
- rag-eval-metrics — Production evaluation metrics (Hit Rate @ K, MRR, NDCG, Reciprocal Rank Fusion) for RAG retrieval benchmarking.
- rag-chunk-visualizer — Interactive Next.js playground for real-time semantic chunking and embedding distance analysis.
- docker-agent-sandbox — Hardened Alpine-based execution container for untrusted LLM-generated code.
- prompt-telemetry-php — Zero-dependency PHP 8.4+ utility for LLM token consumption, latency, and cost telemetry.
- awesome-nfc — Curated NFC tools, hardware guides, and browser-based Web NFC workflows.
Core
AI Engineering
Backend & Data
Cloud & Delivery
