I'm Leah R. Tucker

I design the adoption path and write the playbooks that help developers ship production GenAI on Kubernetes.

About Me

Tech puzzles? They're my kind of fun.

I pick up new systems fast, build the missing pieces end-to-end, and leave documentation that doesn’t punish the next person.

I started in college by reading other people’s HTML source. Still haven’t stopped.

I’ve done this at Stripe, AWS, Sabre, and Avis—then built Blinkt AI, a production RAG system, because I kept seeing the same prototype failures repeat in the wild.

Things end up simpler afterward. That’s the quiet part I chase every time.

Tech Stack

AI & Reasoning Engines

  • Core: PyTorch, Hugging Face, Sentence-Transformers, spaCy
  • Reasoning: Causal Inference (NetworkX), HyDE, RAG, Cross-Encoders
  • Models: OpenAI API (GPT-4o), Fine-tuned Encoders (MiniLM/BERT)
  • Vector Search: Pinecone (GRPC), BERTopic

Backend & Microservices

  • API: Python (FastAPI), AsyncIO, WebSockets (Real-Time)
  • Compute: Docker, Kubernetes (EKS), Celery
  • Data: PostgreSQL (AsyncPG), Redis (Caching/Broker)
  • Languages: Python 3.11+, JavaScript (ES6+)

MLOps & Infrastructure

  • Pipeline: Continuous Fine-Tuning, Automated Data Labeling
  • Cloud: AWS (S3, Boto3), Firebase (Auth)
  • Tooling: Hugging Face Hub, OpenTelemetry, Sentry
  • Monetization: Stripe API (Metering)

Professional Highlights

Stripe

Mapped the onboarding choke points across seven SDK languages → shipped Quickstarts + reference apps that made first success predictable → helped drive +25% Stripe Checkout activation.

AWS

Turned EKS AI/ML adoption into a “golden path” → led the strategy and shipped the canonical best-practices guide with 50+ experts → 8K+ views in 30 days, 92% satisfaction.

Sabre

Made a 2,500-API ecosystem navigable → owned the developer portal experience and publishing workflow so teams could self-serve instead of waiting on governance.

Avis Budget Group (ABG)

Standardized API design through reusable data models + a real API design guide → gave teams a shared language that cut deployment time 75%.