What I work on
APIs and services in Python, TypeScript, and Node.js. I aim for code that reads clearly, fails predictably, and stays easy to operate in production.
AWS infrastructure as code with CDK, CloudFormation, and Kubernetes. CI/CD pipelines and platform tooling that get out of the way of product teams.
Metrics, logs, traces, dashboards, and alerts wired in early, so on-call engineers move from symptom to cause without guessing.
A shareable wish list: save the link, photo, size, color, and notes, then send one public page. Guest lists work from a private manage link; an account keeps several lists in one dashboard. Reservations sit on the shared view so two people do not buy the same thing.
Next.jsTypeScriptPrisma
A chatbot for survivors of gender-based violence. No conversation persistence, no identifiers. The data model is shaped so those features cannot be added later by accident.
PythonLLM
Natural-language analytics: English in, validated read-only SQL out, with the rows and the query shown. Statement whitelisting, planner checks, and row caps sit outside the model.
PythonFastAPISQL
LLM firewalls sit between your application and the model, screening prompts and responses for injection, jailbreaks, and leaked data. How they work, the tools worth knowing, and the attacks that still get past them.
What building Niva, a chatbot for survivors of gender-based violence, taught me about privacy, restraint, and making ethical commitments part of the system design.
A top-down reference for AWS networking. CIDR math, subnets, gateways, NACLs, and security groups, walked through from the outside in across eleven layers.
All writingEmail is the reliable way to reach me. I read everything and usually reply within a couple of days.
ottimothy@gmail.com