Analyze requirements, design AI/NLP solutions.
Proactively contribute ideas to improve system architecture and overall project efficiency.
Build and maintain pipelines that process text documents and support efficient information retrieval.
Collaborate with PM and team members to plan and report on implementation progress.
Integrate LLMs with enterprise data sources (databases, APIs, SaaS platforms) and optimize retrieval and inference performance.
Test and improve pipelines, data processing modules, and LLM integration to ensure reliable, scalable, and high- performance systems.
Job requirement
Bachelor with major Computer Science/ Software engineer
+3 years of experience in AI/NLP Application
Experience fine- tuning or adapting LLMs (LoRA, prompt engineering).
Ability to communicate and collaborate effectively.
Strong understanding of SaaS integration and hybrid system architecture.
Proficiency in Python (3+ years) and SQL.
Experience with vector databases and search techniques (e.g., Weaviate, Pinecone, FAISS, Milvus).
Experience designing and implementing RAG pipelines (e.g., multi- source retrieval, reranking, metadata- driven search).
Experience with text data processing (preprocessing, metadata tagging, feature extraction, embeddings).
Highly motivated to learn and keep up with the latest technologies.
Experience working with GPT- based and open- source LLMs (e.g., OpenAI, LLaMA, Mistral).
Ability to customize and extend low- code platforms (e.g., Dify, N8N).
Experience deploying on cloud platforms (e.g., Azure, GCP, AWS).
Nice to have:
Experience with search and retrieval systems (e.g., Azure AI Search, Elastic, or hybrid search solutions).
Familiarity with LLM evaluation/testing frameworks (e.g., Ragas, Promptfoo).
Understanding of LLM orchestration and product workflows (e.g., multi- step pipelines, chatbot architecture).
Experience with MLOps practices for LLM (model lifecycle, monitoring, drift detection).
Experience with monitoring and logging systems (e.g., Sentry, Grafana).
Experience with CI/CD pipelines (e.g., GitHub Actions).
