Jun Zhou
I'm a dynamic data analyst with a focus on time-series data processing and AI/LLM engineering.
About
I’m a dynamic data analyst with a focus on time-series data processing and AI/LLM engineering. I currently serve as an LLM developer intern at Ecobloom, where I help build domain-specific greenhouse LLMs and multi-agent tools to support energy-efficient and environment-smart decision-making for growers. I enjoy turning high-frequency sensor data into actionable insights and bridging NLP/ML with real-world domain knowledge.
My background spans data engineering, NLP, and AI engineering. I design automated ETL pipelines, extraction layers, and RAG systems; I also develop dashboards and QA workflows to ensure robust, scalable data solutions. I thrive on translating complex data into clear reports and enabling domain experts to leverage powerful AI models.
- 4 years' experience
- Joined March 2026
Skills· 13
Languages
- EnglishFluent
- SwedishAdvanced
- ChineseFluent
Industry experience
- Software & Internet
- Education
- Agriculture & Mining
- Professional Services
Work experience· 4 roles
Designed and implemented automated ETL pipelines to retrieve and preprocess high-frequency time-series data from PostgreSQL databases. Engineered a statistical extraction layer (Mean, StdDev, Range, Trend) to compress raw sensor data, reducing LLM context window pressure by 80% and resolving memory bottlenecks. Leveraged Chain-of-Thought (CoT) and XML-tagging to enhance LLM diagnostic accuracy for greenhouse anomalies; built a Retrieval-Augmented Generation (RAG) system integrating real-time anomalies with greenhouse knowledge, enabling 3 domain-specific LLMs to generate high-fidelity flash reports.
Developed a real-time Audience Response System hub simulator using Python and MQTT, enabling synchronized multi-user data collection with sub-100ms latency for controlled testing. Designed multimodal experiments (30 participants) to evaluate Whisper ASR performance under synthetic speech vs. background noise and assess robustness to human perception.
Collaborated with the tech team to evaluate physical-chemical features and implemented workflow improvements that enhanced product stability validation and increased positive user feedback by 10%. Built 4 QA dashboards in Power BI visualising lab testing workflows across 30+ engine lubricant formulations, enabling real-time progress tracking for commercialisation decisions.
Education
Hire Jun Zhou today
Post a job and invite Jun, or contact them directly. No hidden commission.