Yasaman Asghari
I am Yasaman Asghari, an AI Engineer with 3 years of combined experience in Machine Learning and Generative AI. I currently engineer high-precision RAG systems and optimize inference latency in production.
About
I am Yasaman Asghari, an AI Engineer with 3 years of combined experience in Machine Learning and Generative AI. I currently engineer high-precision RAG systems and optimize inference latency in production.
I have a strong foundation in model training, data preparation, and ML algorithm evaluation, transitioning from research to building scalable AI solutions. I am skilled in Python and selecting the right LLMs for business needs.
- 6 years' experience
- Joined December 2025
Languages
- EnglishAdvanced
Industry experience
- Software & Internet
- Professional Services
More experts in Utrecht, Netherlands
Work experience· 4 roles
AI Engineer on the Continuous AI Delivery platform. Engineered high-precision RAG pipelines using LangChain with query decomposition and cross-encoder rerankers to improve retrieval accuracy and context relevance. Implemented semantic caching and context-aware optimizations, reducing latency for recurring queries from seconds to under 1s, and parallelizing processing with Python Async/Await to cut complex query execution time by 40%. Developed production-ready GenAI use cases (Conversational AI, Summarization) using the OpenAI API and refined prompts for consistent tone and factual accuracy. Evaluated model options to balance cost, capability, and inference speed; collaborated on software tasks including Git and Docker deployments.
Self-directed research and upskilling period in the Netherlands. Transitioned from traditional ML to AI Engineering by mastering LLMs, RAG architectures, and vector search technologies. Built and tested prototype Q&A pipelines with Hugging Face Transformers to compare embedding models, analyzing trade-offs between retrieval speed and accuracy. Wrote Python scripts to clean and chunk unstructured text for vector database ingestion (ChromaDB), focusing on data quality and context preservation. Conducted model benchmarking on summarization tasks and logged performance metrics to understand cost-versus-performance trade-offs.
Developed and trained time-series forecasting models using Scikit-learn; applied hyperparameter tuning (GridSearch) to optimize parameters, achieving an 18% improvement in forecast accuracy. Engineered feature engineering from raw journey logs with Pandas to improve robustness of sales prediction models. Established a rigorous evaluation framework using RMSE and MAE to compare multiple algorithms for production use. Built automated ETL workflows using Apache Airflow to ingest and clean heterogeneous data into MySQL, reducing manual data preparation time by 30% and ensuring high-quality training datasets.
Education
Certifications and awards
Hire Yasaman Asghari today
Post a job and invite Yasaman, or contact them directly. No hidden commission.




