7 years writing developer documentation for APIs, SDKs, and complex technical systems. Currently expanding into ML and AI platform documentation — applying the same rigor to model workflows, optimization tooling, and framework integration that reduced support tickets by 58% at Chargebee.
Documentation projects where the outcome was a business metric, not a page count.
ML engineers converting PyTorch models to ONNX frequently hit underdocumented edge cases around dynamic axes, opset versioning, and quantization-aware training compatibility. Designed and wrote a developer-facing guide covering the full export workflow: model preparation, dynamic shape configuration, validation with ONNX Runtime, and post-export INT8 quantization. Structured for two audiences — engineers running first-time exports and those optimizing for edge deployment. Includes runnable Python code blocks and a troubleshooting reference for common conversion errors.
Chargebee's subscription API had grown to 160+ endpoints over 5 years with no single documentation owner. State transition logic was buried in a PDF, error codes had no descriptions, and support was fielding 300+ developer questions per month. Audited every endpoint, ran discovery sessions with 5 engineers, and rebuilt the reference in OpenAPI 3.1 with embedded Node, Python, Ruby, PHP, and Go samples. Shipped in 9 weeks.
Analytics showed new developers were averaging 3.2 days from signup to first successful API call. Ran 6 usability sessions with developers new to Chargebee, mapped every drop-off point, and designed a stack-adaptive quickstart (Node, Python, Ruby, PHP, Go) with inline sandbox credentials and runnable Postman collections. Time-to-first-call dropped to under 4 hours post-launch.
Internet Brands operated 6 product documentation sites with no shared standards — terminology, voice, formatting, and code conventions varied wildly across legal and healthcare platforms. Researched 4 industry style guides (Google, Microsoft, Apple, Divio), ran a content audit across all 6 products, and built a modular style guide adopted by 4 product lines. Reduced review cycles by 40% by resolving style debates before they happened.
Available formats: API reference, quickstart guides, conceptual docs, ML platform overviews, SDK references, error references, changelog writing, and style guide documentation. All samples include the brief, review cycle, and outcome metrics on request.
Open to senior remote technical writing roles — ML and AI platform documentation, API developer experience, and docs infrastructure. Based in Downey, CA. Available immediately.