Articles

Lessons so far

What a co-op term on an ML localization project has taught me.
Keeping suggestions live

Latency budgets and caching for autocomplete that has to feel instant.
Choosing a model family

How I compared n-grams, LSTMs, autoencoders and transformers for the prototype.
Working in the open

Building on a public platform with a large volunteer community.
Evaluating five million strings

What the corpus evaluation actually involved, and what it taught me.
Prototyping localization autocomplete

Building an autocomplete prototype for Pontoon, and what it took to get there.
Joining Mozilla

My role on the team behind Pontoon, Mozilla's localization platform.

No matches.

Press Esc or click anywhere to close