Where I have worked and what shipped.

Financial NLP research: event-driven pipelines that turn Federal Reserve communications into structured trading signals.

• 200K+data pointsBuilt a fault-tolerant, event-driven pipeline in Python and FastAPI on AWS processing 200K+ financial data points in real time with zero data loss under variable load.
• 99.9%availabilityDesigned a distributed ingestion system on AWS Lambda and S3 covering 3.7K+ time-series intervals with schema validation, sustaining 99.9% availability.
• <300msP99 latencyArchitected REST APIs in FastAPI and DynamoDB exposing structured financial data across microservices at sub-300ms P99 latency.
• 97%faster deploysImplemented Docker-based CI/CD on AWS with Jenkins, cutting deployment time by 97%, and added Prometheus/CloudWatch observability to catch defects before release.

• Python
• FastAPI
• AWS Lambda
• DynamoDB
• S3
• Docker
• Prometheus

Medical imaging ML: a diagnostic pipeline that cut clinical processing from twenty minutes to under five.

• 90%accuracyBuilt an end-to-end ML pipeline in PyTorch on GCP reaching 90% classification accuracy across 1,250+ patient cases.
• 4xthroughputReduced clinical processing time from 20+ minutes to under 5, a 4x throughput improvement for the reviewing team.
• 10TB+imaging dataProvisioned fault-tolerant GCP infrastructure with Terraform and Firebase for real-time inference across 50K+ scans (10TB+) with zero-loss data handling.

• Python
• PyTorch
• GCP
• Terraform
• Firebase

Three years on distributed backend systems for Fidelity Investments and BT Group — event-driven ETL, serverless platforms, and latency work at 100K+ customer scale.

ClientsFidelity Investments, BT Group

• 10+systems integratedArchitected an event-driven ETL system for Fidelity Investments in Java, Spring Boot, and Kafka, automating integration across 10+ enterprise systems at 99% uptime with full test coverage on production paths.
• 3K+customers onboardedOwned a serverless platform in Python, Node.js, and AWS Lambda handling async workflows and onboarding 3K+ customers through REST integrations — including reliability and on-call.
• 40%P99 reductionCut P99 latency by 40% and eliminated Sev1 incidents by rearchitecting the Redis caching and Elasticsearch layers of a system serving 100K+ customers.
• <200msP99 latencyDelivered production REST and GraphQL APIs in Java, Spring Boot, and MySQL for BT Group, processing 1,000+ daily transactions at sub-200ms P99.

• Java
• Spring Boot
• Kafka
• AWS Lambda
• Redis
• Elasticsearch
• MySQL
• GraphQL

Secrets management and telecom microservices on AWS.

• 8+services securedDesigned and owned a centralized secrets management microservice in Java, Spring Boot, and MongoDB, securing cross-service auth across 8+ microservices with OAuth 2.0 and JWT and zero credential breaches in production.
• 10K+daily transactionsBuilt telecom microservices processing 10K+ daily transactions with comprehensive JUnit coverage and zero production regressions.

• Java
• Spring Boot
• MongoDB
• AWS
• OAuth 2.0
• JWT