Looking for a job

M Joshua

@mjoshua

Senior AI/ML Engineer at Spring Health building clinical RAG and multi-agent LLM systems that improved retrieval accuracy by 35%.

United States

Message

At Spring Health, I architected clinical knowledge retrieval with production RAG pipelines on FAISS and Pinecone, improving retrieval accuracy by 35% while serving queries at sub-200ms p95 latency.

I designed multi-agent LLM workflows for mental health intake that reduced clinician intervention by 40% and administrative workload by 25%. I also fine-tuned Llama 3 and Mistral with LoRA and QLoRA on de-identified clinical datasets, gaining 28% in contextual accuracy.

At Google, I built and deployed TensorFlow models and production ML pipelines powering real-time analytics across internal products. I also designed NL-to-SQL pipelines and helped bring pretrained transformer models and vector search into use across partner teams.

Experience

Work history, roles, and key accomplishments

Current

Senior AI/ML Engineer

Current

Spring Health

Jan 2022 - Present (4 years 9 months)

Architected production RAG pipelines and multi-agent LLM workflows for clinical knowledge retrieval, improving retrieval accuracy by 35% and reducing clinician intervention by 40%. Fine-tuned open-source LLMs with LoRA/QLoRA and built HIPAA-compliant AI services with MLOps pipelines on AWS SageMaker.

Python SQL PyTorch TensorFlow Hugging Face Transformers Lang Chain LangGraph React RAG Faiss Pinecone LoRAs QLoRA GPT 4 LLaMA 3 Mistral AWS Sagemaker Docker Kubernetes MLFlow HIPAA PII Masking RBAC A B Testing Langsmith

Data Scientist

Google

Jan 2017 - Dec 2021 (4 years 11 months)

Built and deployed TensorFlow models on Google Cloud AI Platform, powering real-time analytics across internal products. Engineered production ML pipelines with TFX, Kubeflow, and Apache Beam, and drove adoption of transformer models and vector search.