Minghan (Max) Zhu

Building ML and data infrastructure for adaptive DBS research, from neural signal processing to clinician-reviewed parameter-selection workflows.

• Design and maintain pipelines for implanted neural recordings, wearable motor data, sleep measures, clinical assessments, and device telemetry.
• Develop ML modules for adaptive DBS research, including stimulation simulation, probabilistic biomarker response modeling, surrogate LFP generation, symptom-profile metrics, and constrained optimization.
• Build clinician-facing visualization and analysis tools that support review of adaptive stimulation parameters and experimental outcomes.

• Adaptive DBS
• Neural devices
• Signal processing
• Clinical ML

Co-founded an AI pet-tech startup funded by the Eugene M. Lang Center.

• Designed and implemented a custom ERP system for pet product manufacturers, integrating with CRM and inventory management systems.
• Reduced manual data entry time by 67% through workflow automation.
• Used SDXL, ControlNet, Shap-E, and agentic LLMs to build a semi-supervised prototype pipeline for custom pet apparel design.

• ERP
• LLMs
• SDXL
• ControlNet
• Supply chain

Built LLM-powered search and due-diligence software for vertical research.

• Designed and built an enterprise-grade LLM search engine with a Next.js frontend and Redis caching for high-speed results.
• Developed an agent-based copilot for financial due diligence with real-time analysis of filings, reports, and social media.
• Helped triage leads and generate investment insights through research automation.

• LLM search
• Next.js
• Redis
• Agents
• Finance

Created housing data pipelines and analytics tooling for carbon analysis.

• Developed Airflow pipelines for municipal housing data ingestion, expanding coverage and improving accuracy.
• Built a web portal for US housing carbon analysis with Laravel, PHP, and React.
• Reduced manual entry time by 37%.

• Airflow
• React
• Laravel
• Housing data

Improved healthcare data ingestion, reconciliation, and support diagnostics.

• Redesigned Cedar's SFTP data puller with Python, Django, PostgreSQL, and Airflow.
• Consolidated 43 providers, patched an ETL vulnerability, and built HIPAA-compliant pipelines for real-time patient record reconciliation.
• Standardized alerting runbooks and reduced support engineering time by deduplicating AWS CloudWatch and Sentry diagnoses.

• Healthcare data
• Python
• Django
• Airflow
• PostgreSQL

Built delivery analytics, dashboard workflows, and menu-mapping automation.

• Developed a dynamic client dashboard with Angular and Blazor, integrating third-party delivery APIs and real-time analytics.
• Improved client retention by 23.5%.
• Designed semantic mapping tools in Python using Word2vec and fastText to reduce manual menu mapping by 66.2%.

• Angular
• Blazor
• NLP
• Delivery APIs

Co-invented a semantic catalog mapping system later issued as a patent.

• Co-invented a semantic-based catalog data mapping system filed on December 30, 2020 and issued on June 30, 2022.
• Patent: US20220207507A1.

• Patent
• Semantic mapping
• Catalog data