Joshua

Key Skills

Software

Don't disclose

Top Subject Matter

GenAI/LLM-powered semantic search

RAG evaluation

and responsible AI readiness for e-commerce/marketplace discovery

Top Data Types

Text

Document

Top Task Types

Fine-tuning

Freelancer Overview

Senior Machine Learning Engineer & Data Scientist — DoorDash (Dec 2022–Present) focused on LLM evaluation and agentic AI. Brings 12+ years of professional experience across legal operations, contract review, compliance, and structured analysis. Core strengths include Don't disclose and Hugging Face Transformers. Education includes Master of Science, Georgia Institute of Technology (2021) and Bachelor of Science, University of Georgia (2015). AI-training focus includes data types such as Text and labeling workflows including Evaluation, Rating, and Fine-tuning.

Labeling Experience

Senior Machine Learning Engineer & Data Scientist — DoorDash (Dec 2022–Present) focused on LLM evaluation and agentic AI workflow evaluation

Don't discloseText

Built and deployed LLM-as-a-Judge evaluation frameworks to detect hallucinations, validate semantic correctness, and benchmark model performance before release. Performed offline/online evaluation of LLM-powered semantic search and RAG systems to support safe, observable deployments. Coordinated evaluation and release readiness across multi-agent workflows that enforce structured outputs and grounded generation. • Hallucination detection and semantic correctness validation for production LLM releases • Benchmarking multiple models (e.g., GPT-4/Claude) for quality/latency/token-efficiency tradeoffs • Offline/online evaluation of RAG pipelines using embedding-based retrieval • Structured validation layers and guardrails integrated into evaluation and agent execution

2022 - Present

Machine Learning Engineer — Clover Health (Aug 2019–Nov 2022) Transformer-based clinical NLP training work

TextFine-tuning

Fine-tuned transformer models for clinical NLP tasks using task-specific objectives with PyTorch and Hugging Face Transformers. Developed embeddings-based feature extraction workflows from physician notes to support downstream chronic condition detection and clinical review automation. Established reproducible training and evaluation pipelines across claims, EHR, and NLP data sources. • Fine-tuning ClinicalBERT/BioBERT-based models for clinical classification tasks • Embedding generation from physician notes for feature extraction • Training/evaluation workflow reproducibility across EHR and claims inputs • Development of clinical summarization and information extraction prototypes using transformers

2019 - 2022

Education

Work History