Datamuse is a tiny software venture born in the early days of the Web and still stubbornly stuck there. We make brainstorming tools like RhymeZone, OneLook, and the Datamuse API, that quietly reach more than 10 million writers every month. (Here's a recap of some of the projects we've been working on during the past few years.)
We’re looking for contract programmers — computer science students or seasoned software developers — to help us improve our sites in 2022 and build entirely new ones. We hire contractors on a project basis and will consider longer-term engagements if there's mutual interest. Do you have experience in, or enthusiasm for, natural language processing, machine learning, or data visualization? Please read the project descriptions below, and if any of them pique your interest, give us a sense of your experience, time frame, and desired terms/rate in the feedback box at the bottom of this page. Thank you!
Research
(Filled) OneLook Thesaurus language modeling: This project is suited for someone studying natural language processing and interested in applying their skills to improve a live Web application. You will experiment with using large language models such as GPT-3 to increase the quality and breadth of the results returned for certain kinds of queries made to OneLook Thesaurus. You will be responsible for implementing and evaluating your solutions with data resources that we provide. Project scope: 40+ hours over 3 months (open-ended); Desired skills: Proficiency in Python; experience with processing text data; machine learning basics.
OneLook Thesaurus internationalization: This project is suited for someone studying natural language processing and interested in applying their skills to improve a live Web application. You will be responsible for implementing and evaluating your solutions with data resources that we provide, and seeking out additional data resources in the target language. Project scope: 40+ hours over 2 months (open-ended); Desired skills: Proficiency in Python; experience with processing text data; machine learning basics; proficiency in Spanish, French, or Mandarin Chinese.
Advanced NLP mini-projects: We have several ongoing research efforts aimed at making our brainstorming tools better. We apply the latest findings from the NLP and ML research communities to the tasks of clustering, filtering, ranking, annotating, and visualizing the words, phrases, and ideas that we suggest to writers. Example problems include word sense induction; generating illustrative example sentences for dictionary words using large language models; ranking words in multiple languages by different aspects of style; and evaluating thesaurus quality. If you have NLP experience and a strong interest in this problem domain, we'd love to engage with you on one of these projects. Desired skills: Significant research-oriented NLP experience; proficiency in Python and with pytorch / huggingface tools; experience with processing text data. Project scope: Approximately 10 of these projects are of interest currently, each one needing 80-160 hours of focused attention; inquire within for more details. We're eager to evaluate and publish any new findings that arise from this work at academic conferences, and can collaborate with the contractor on the publication process if there is mutual interest. (College co-op/internship possible if applicable.)
Spruce: We're working to make our Spruce brainstorming tool more comprehensive, more accurate, and faster. For this project you will work with our back-end semantic search and indexing code to scale Spruce to find connections between any piece of writing and hundreds of millions of documents. Desired skills: Proficiency in Python; some experience with large-scale data pipelines; data analysis experience.
Front-end development
(Filled) Brainstorming viz: For this project, a web developer familiar with D3.js or similar visualization frameworks will implement a graphical output mode for OneLook Thesaurus wherein similar terms are grouped into clusters and related clusters are linked to each other in an interactive visualization. Project scope: 80 hours over 2 months (design + implementation of prototype). Desired skills: Proficiency in web design and/or development; ReactJS.
(Filled) Google Docs add-on: For this project, a developer familiar with Google Apps Script will improve the quality and reliability of our Google Docs add-on. Project scope: 40+ hours. Desired skills: Some familiarity with Google Apps Script and past experience deploying similar extensions or add-ons.
(Filled) New microsite: For this project, a front-end web developer or designer will help us get a new website off the ground by designing, prototyping, and implementing an application around back-end APIs that we provide. Project scope: 40+ hours. Desired skills: Proficiency in web design and/or development; ReactJS.
RhymeZone visual redesign: You might call this project "Mess with success". On one hand, RhymeZone is a rare 1990s-era website that is still going strong, and we'd like to keep it that way: most of our attempts to modernize the layout over the years have been met with jeers rather than cheers. On the other hand, we know that the site can be much stronger, both in terms of its feature set and its user experience, especially with all of the work we've done on the API in the past few years. As a start we'd like to partner with a web designer to bring the design into the 2020s without harming page speed or other aspects of usability. Project scope: 480 hours. (College co-op/internship possible if applicable.) Desired skills: Proficiency in web design; ideally, some experience focused on designing for high-traffic information tools.
Infrastructure
Java performance tuning: We're looking for a Java server performance expert to help us reduce the latency of the Datamuse API. You will use tools such as CodeGuru to profile live and synthetic query loads, and implement and test changes to our API codebase accordingly. Project scope: 40 hours. Desired skills: Up-to-date Java expertise.
(Filled) Self-hosted LLM: We're looking for a systems-oriented
machine learning engineer to help
us deploy large language models suited for our applications on our
own cloud resources. This project will involve evaluating various open source
distilled large language models on our own benchmarks and deploying them
in our cloud, optimizing for a combination of accuracy, latency, and serving cost.
Project scope: 40+ hours.
Desired skills: Experience deploying large-scale machine learning models.
Data engineering for i18n: We'd like to make our services available in as many languages as possible, but our entire stack (our data processing scripts especially) is not well abstracted to allow us to add new languages easily. We'd love to work with a software engineer proficient in Python data processing pipelines to help us achieve this, applying modern software engineering practices to rewrite parts of our stack with internationalization in mind. Project scope: 480 hours (College co-op/internship possible if applicable.) Desired skills: Python proficiency, ideally with pipelines that process large amounts of textual data. Attention to detail.
Crowd-sourcing automation: We plan to run a few large-scale evaluations and annotation tasks related to new and current Datamuse features, and we're looking for a software engineer to automate some of the data collection and analysis steps. Project scope: 80+ hours over 2 months; Desired skills: Proficiency in Python; familiarity with Amazon Mechanical Turk or similar crowd-work platforms.
Product
Product marketing generalist: We're looking for an enthusiastic word nerd to help us explain our services to the millions of writers who already use our sites, and millions more in the future. The ideal contractor will be flexible enough to do all of the following: design and record screencasts; write compelling App Store descriptions; oversee creation of graphic design assets; use Google Analytics wisely; run user studies in which you talk to real humans; design, deploy, and analyze surveys; suggest user experience improvements and new features; oversee translation/localization of the help text in our products; and invigorate a nearly dormant social media account. If you're interested in product marketing and you're looking to have a small impact on a huge company, then this is totally not the opportunity for you! But if you're looking to have a huge impact on a tiny venture, then please inquire within. Project scope: ~480 hours over 4 months. (College co-op/internship possible if applicable.)
Do any of these contract projects pique your interest?
Please give us a sense of your experience, time frame, and desired terms in the feedback box below.
Some praise for our OneLook Thesaurus project! (More)