Machine Learning Engineer#
National Artificial Intelligence Center (NAIC) · naic.az Feb 2025 – Present
- Built a query processing pipeline using quantized DistilBERT for NER and query classification; deployed to AWS.
- Converted SBERT and DistilBERT to ONNX; applied dynamic quantization, reducing model size by 39% with negligible accuracy loss.
- Applied mixed-precision training to accelerate fine-tuning while maintaining benchmark parity.
- Implemented cross-encoder LLM re-ranking and a Lightweight Re-ranking Transformer (LRT) for precision stage.
- Ran systematic latency vs. accuracy benchmarks across retrieval + re-ranking pipeline stages.
Junior Machine Learning Engineer#
National Artificial Intelligence Center (NAIC) · ailab.az Jun 2023 – Feb 2025
- Designed and deployed a hybrid retrieval system (BM25 + dense vectors) fine-tuned on 500k+ domain-specific documents.
- Defined evaluation suite (MRR@10, NDCG@5, precision-recall) and achieved 85% accuracy in user testing.
- Integrated Elasticsearch, FAISS, and Transformer encoder models into a unified retrieval stack.
- Optimized data ingestion pipelines with parallel processing, achieving 4× throughput improvement.
- Applied NLP preprocessing (tokenization, stopword removal, entity normalization) to improve retrieval quality.