Open to opportunities

Derrick Lee

@derricklee1

Message

Senior Machine Learning & AI Engineer specializing in MLOps and agentic GenAI systems.

United States

Message

What I'm looking for

I’m looking for a senior role where I can ship production-grade agentic GenAI and MLOps—building guardrails, evaluation, and governance so models reliably drive measurable business outcomes across channels.

I’m a Senior Machine Learning & AI Engineer with 10+ years designing and operating production-grade ML systems. My focus is end-to-end delivery—from distributed GPU training and evaluation through deployment, observability, and governance-ready outcomes.

I’ve led delivery pods for Fortune 100 enterprises (including Verizon and Boeing), translating ambiguous business problems into measurable AI products. Recently, I architected a multi-agent personalization platform targeting under-200ms latency with robust stateless-but-stateful orchestration, plus guardrails to drive low hallucination and high tool-call success rates.

I also build the measurement and rollout machinery that makes GenAI reliable at scale. From KPI frameworks and staged traffic governance to experiment tagging and rollback-by-config protocols, I make agentic systems auditable and safe for real operations.

Across industries, I’ve delivered agentic RAG assistants, supply-chain planning copilots, real-time personalization, and cybersecurity ML—plus enterprise agentic code review integrations. I enjoy the hard parts: evaluation harnesses, cross-cloud routing, secure data access, and production hardening that turns models into dependable products.

Experience

Work history, roles, and key accomplishments

Current

Senior Machine Learning Engineer

Current

Blend360

Jul 2025 - Present (11 months)

Architected Verizon Next Best Conversations agentic personalization, delivering a multi-agent orchestrator and guardrails with <200ms end-to-end latency and target <2% hallucinations. Also led AI data onboarding & audit automation for a national RIA aggregator and built an agentic AI code-review integration for Azure DevOps with CI/CD hardening and evaluation.

Multi agent orchestration MCP (Model Context Protocol)Vertex AI fastAPI OpenTelemetry Azure DevOps

Lead Machine Learning Engineer

Tredence

Apr 2024 - Jun 2025 (1 year 2 months)

Led a Fortune 100 grocery retailer agentic merchandising co-pilot (280+ seats), using LangGraph multi-agent planning and tool access over Databricks Unity Catalog + Snowflake, with an evaluation harness gating releases on ≥92% agreement vs. category managers. Built production GenAI assistants (SAP IBP supply-chain exception copilot; Azure Teams contract/NDA co-pilot) and modernized MLOps on Kubern

Databricks Unity Catalog Snowflake Kubernetes VLLM Azure Databricks

Senior ML & GenAI Engineer

SoluLab

Oct 2022 - Mar 2024 (1 year 5 months)

Architected an AI + blockchain fractional real-estate tokenization platform, combining XGBoost/LightGBM valuation with ERC-1400 tokens on Polygon and Chainlink-orchestrated rent distributions, eliminating manual reconciliation for 2,400+ token holders. Built GenAI co-pilots for legal and DeFi fraud/wash-trading (RAG over 80k contracts, Pinecone; Solidity drafting with Slither/MythX; GNN detection

XGBoost LightGBM RAG With Pinecone JSON Schema Constrained Decoding LangGraph PyTorch Geometric GNN Kafka Streams

Senior Machine Learning Engineer

Sigmoid

Jan 2019 - Sep 2022 (3 years 8 months)

Built a real-time personalization engine using two-tower retrieval + LightGBM re-ranking and a contextual bandit policy, increasing session conversion +6.2% vs. a collaborative-filtering baseline. Engineered Feast feature pipelines, Kafka/Spark streaming at 180k events/sec, and led Bayesian hierarchical MMM/MTA that supported redirecting ~$38M of annual media spend.

Two Tower Retrieval LightGBM Contextual Bandits TensorFlow Databricks Delta Lake Feast Kafka Spark Structured Streaming Bayesian MMM (PyMC)Shapley Value Attribution

Machine Learning Engineer

SparkCognition

Aug 2016 - Dec 2018 (2 years 4 months)

Developed DeepArmor AI for endpoint cybersecurity, creating malware detection ensembles on PE files and deploying a retraining pipeline on Spark/EMR to keep false positives <0.01%. Built predictive maintenance models (LSTM/RUL with hybrid anomaly detection) and production feature stores on Kafka/Cassandra for millisecond-class inference, plus contributions to renewable energy forecasting within Au

PE Malware Feature Engineering PyTorch Kafka Cassandra