Teddy – React, Python, AWS

As Lead for Infrastructure & Data at Phare Health, Teddy owned the architecture and delivery of production clinical-data infrastructure supporting ingestion, de-identification, analytics, and AI workflows. He worked closely with healthcare and domain experts to translate clinical, privacy, and operational requirements into scalable systems, including distinct data paths for real-time inference and model-training workloads.

• Led architecture and delivery across clinical-data ingestion, de-identification, AI/data workflows, and AWS infrastructure.
• Defined separate architectures for live inference and model-training workflows, balancing latency, privacy, reliability, and downstream data needs.
• Built a GPU-accelerated de-identification pipeline using OCR and transformer-based PHI detection.
• Delivered a production de-identification platform processing 20k+ clinical encounters daily, with data-quality checks, provenance, traceability, and observability built into the pipeline.
• Designed fault-tolerant workflows across HL7/FHIR, Kafka/MSK, Airflow/MWAA, Lambda, SQS, and S3.
• Worked directly with healthcare/domain experts to shape requirements, validate technical decisions, and ensure solutions reflected real clinical and business workflows.
• Communicated architecture, risks, and technical trade-offs clearly to both technical and non-technical stakeholders.

Embedded with The Weather Company through Andela, Teddy led a team of 4 building AI-powered analytics and visualization tools that connected weather patterns with external business datasets, including Nielsen sales data. The work focused on turning complex machine-learning outputs into insights that business stakeholders could understand and act on.

• Led a team of 4 engineers delivering AI-powered analytics and visualization features from design through production.
• Built AI Explainers using prompt engineering to translate machine-learning insights into clear, actionable narratives for non-technical stakeholders.
• Designed and built the evaluation pipeline for the AI Explainers, establishing repeatable quality criteria and feedback loops that guided prompt/model iteration and improved reliability before production rollout.
• Worked with product and business stakeholders to understand how weather data and external commercial datasets, including Nielsen sales data, could be combined to support better decision-making.
• Built NLP pipelines using Transformers and PostgreSQL for large-scale text de-identification and summarization.
• Designed and delivered interactive Mapbox and D3 dashboards for exploring complex weather and business datasets.
• Improved engagement with analytics experiences by approximately 25% through more intuitive visualizations and explainable AI outputs.
• Helped shape technical approaches where requirements involved ambiguous analytical questions rather than predefined implementation specifications.
• Communicated AI outputs, limitations, evaluation results, and technical trade-offs clearly across engineering, product, and business stakeholders.

At BlockBar, Teddy worked in a small product team building a luxury digital-asset marketplace spanning e-commerce, blockchain, and NFTs. He owned full-stack product delivery across customer-facing workflows and fully owned the design, implementation, and production operation of an AI/RAG customer-support solution that helped the team scale support during high-volume product drops.

• Fully owned a RAG-based customer-support solution end-to-end, from identifying the business problem through architecture, implementation, deployment, monitoring, and iteration.
• Built retrieval and grounding workflows over product/domain knowledge, with human escalation for complex or low-confidence cases.
• Defined and ran evals for retrieval relevance and response quality, and iterated on prompts, retrieval strategy, escalation thresholds, latency, and reliability.
• The solution automatically handled approximately 80% of customer issues, substantially reducing support and engineering on-call overhead during high-volume product drops.
• Worked closely with founders/product stakeholders to prioritize customer problems and make product decisions based on business impact.
• Built full-stack marketplace features across backend APIs, frontend experiences, data, integrations, and AWS infrastructure.
• Delivered a real-time sales lobby that improved the bundle-purchase experience and increased sales volume by 15%.

For Meta, Teddy built features for a large-scale internal support platform used to manage support workflows more efficiently. He worked across backend and frontend systems on platform migration, privacy tooling, and operational improvements that made the system more reliable and easier to maintain.

• Led end-to-end feature development across product and platform workflows for a large-scale internal support system.
• Led discovery and solution design for an ambiguous privacy initiative that needed to give developers access to user data for testing and debugging without enabling misuse.
• Interviewed stakeholders and domain experts, reviewed previous proposals and white papers, and translated the findings into a design document with architecture, milestones, timelines, and implementation options.
• Delivered a zero-downtime migration of core ticketing workflows to a modern internal stack.
• Reduced support tickets by 30% and maintenance time by 50% through platform reliability and workflow improvements.
• Improved privacy tooling and operational debugging success by 31%, reducing dependence on internal support teams.
• Partnered across product, design, and engineering to ship workflow improvements that increased system stability and operational efficiency.

At Microsoft, Teddy worked on Azure AD / Entra ID enterprise provisioning, partnering directly with SaaS companies including Slack, Zoom, Limble, and ServiceNow. The role was highly client-facing: understanding partner architectures, troubleshooting production provisioning issues, shaping integration approaches, and coordinating across customer and Microsoft engineering/product teams.

• Worked directly with engineering teams at enterprise SaaS companies including Slack, Zoom, Limble, and ServiceNow on Azure AD/Entra ID provisioning integrations.
• Acted as a technical partner during complex production issues, diagnosing SCIM/provisioning failures and determining whether changes were required in the customer integration, Microsoft platform, or both.
• Coordinated across partner engineering teams and Microsoft product/engineering stakeholders to drive issues through to production resolution.
• Built and improved SCIM-compliant provisioning infrastructure supporting enterprise identity workflows at scale.
• Helped improve automation, reliability, developer experience, and troubleshooting across provisioning integrations.
• Contributed to approximately 45% growth in provisioned applications, ~30% fewer support tickets, and 95% partner satisfaction.
• Regularly translated complex identity and distributed-system issues into clear recommendations for both technical and non-technical stakeholders.

Embedded with Stem Disintermedia through Andela, Teddy worked directly with engineering, sales, and business-intelligence stakeholders on a music-data platform processing data from Spotify, YouTube, iHeartRadio, and other major platforms. He helped shape and deliver large-scale data infrastructure, replacing brittle workflows with observable, production-grade systems.

• Built high-volume data pipelines ingesting about 1 billion records per day into BigQuery.
• Replaced an unstable, unobservable collection of shell scripts and cron jobs with Apache Airflow workflows deployed on Kubernetes.
• Scaled ingestion from approximately 50 million to billions of records per day from Spotify, YouTube, iHeartRadio, and other music platforms.
• Worked directly with sales, business intelligence, and engineering teams to gather requirements and design the replacement architecture.
• Introduced workflow observability, dependency tracking, restartability, and historical backfills.
• Used Apache Avro to compress data in transit and storage, reducing network and infrastructure costs.
• Improved analytics efficiency by making large-scale data more reliable and accessible for downstream teams.
• Automated external integrations to improve supply chain reliability and reduce manual operational work.
• Worked on cloud-native data infrastructure using Airflow, Docker, Kubernetes, and Avro.

Palladium Group is a large global consultancy, where Teddy worked on national-scale digital health and interoperability programs covering 900+ healthcare facilities. This involved working across client/program stakeholders, healthcare domain experts and engineering teams to translate operational needs into systems that could work reliably at scale.

• Worked within Palladium, a large global consultancy, across client/program stakeholders, healthcare domain experts, and technical teams.
• Helped shape and deliver a national-scale healthcare interoperability layer supporting 900+ facilities.
• Translated healthcare and operational requirements into integration architectures and production workflows across heterogeneous clinical systems.
• Helped improve patient identification and reduce duplicate records across distributed healthcare environments.
• Worked directly with stakeholders throughout discovery, implementation, rollout, and production support.
• Communicated technical constraints, architecture decisions, and trade-offs to both technical and non-technical stakeholders.