About the Role
We are looking for a software engineer/creative technologist with 3-5 years of experience who can move comfortably between prototype and production.
You will help Gooey.AI build experimental and core platform features that allow users to upload multimodal datasets, generate embeddings, visualize semantic relationships, evaluate model behavior, discover meaningful clusters, and create new AI workflows from latent-space insights.
This is a role for someone who enjoys building real systems, but is also excited by open-ended creative and research questions: How do we make a vector space visible? How do we help a non-technical artist understand why two images, poems, sounds, or archival objects are related? How do we evaluate whether an embedding model is useful for humanities research? How can latent-space arithmetic reveal unexpected relationships? How can a researcher move from “I found an interesting cluster” to “I built a reusable AI workflow others can fork”?
You will work closely with Gooey.AI’s product, engineering, and creative leadership, as well as artists, researchers, faculty, and cultural partners.
What You’ll Work On
Multimodal latent-space visualization: Build interactive web-based tools for exploring multimodal vector spaces. This may include constellation-style maps of image, text, audio, and video datasets; cluster views; similarity paths; semantic neighborhoods; hover previews; search overlays; and ways to move between 2D/3D visualization and underlying high-dimensional embeddings.
Embedding model integration: Integrate new embedding models into Gooey.AI, including models from Google, OpenAI, Jina, Meta, AWS, and open-source providers. Build clean abstractions so workflows can compare and swap embedding models without locking users into a single provider.
Evaluation and benchmarking: Develop ways to evaluate embedding models for artistic, cultural, humanities, and multilingual use cases. This may include golden datasets, similarity judgments, cross-modal retrieval tests, clustering quality, low-resource language performance, explainability checks, and human-in-the-loop evaluation workflows.
Dataset exploration and discovery: Build tools that help users find interesting trends, outliers, neighborhoods, bridges, gaps, and clusters in cultural datasets. The goal is not only to retrieve the “right” result, but to help users discover unexpected and meaningful relationships.
Latent-space reasoning and arithmetic: Prototype methods for exploring relationships between embeddings: analogies, differences, transformations, semantic directions, interpolations, and “A is to B as C is to…” style experiments across text, image, sound, and other modalities.
Generative workflows from latent-space insight: Explore how the latest AI models can generate images, sounds, text, or other artefacts from latent-space representations, retrieved clusters, interpolations, or semantic transformations. Help turn these experiments into reusable Gooey.AI workflows that artists and researchers can adapt.
Agentic interfaces for non-technical users: Help build agentic interfaces that allow users to describe their intent in natural language, prepare datasets, choose models, run retrieval and clustering workflows, inspect results, generate outputs, and publish reusable workflows without needing to write code.
Core Gooey.AI platform improvements: Contribute to the underlying platform services that make this work scalable, reliable, and reusable: APIs, model integrations, vector databases, workflow execution, evaluation infrastructure, deployment tools, logging, analytics, documentation, and open-source releases.
Example Projects
You might work on:
A web-based “semantic constellation” that visualizes 1,000–100,000 images using embeddings, dimensionality reduction, and interactive thumbnails.
A workflow that lets a researcher compare Gemini, Jina, CLIP, ImageBind, Nova, and other embedding models on the same cultural dataset.
A tool that finds clusters, outliers, bridges, and surprising relationships in an archive of images, poems, sounds, and research notes.
A latent-space arithmetic playground for exploring analogies and transformations across humanities datasets.
An evaluation workflow where artists and researchers rate whether discovered relationships are meaningful, surprising, useful, or misleading.
An agent that helps a non-technical researcher upload a dataset, choose models, build an exploration workflow, interpret the results, and publish the workflow for others to fork.
A generative pipeline that turns latent-space discoveries into new images, soundscapes, prompts, exhibitions, or research artefacts.
Technical documentation and example notebooks/workflows that make this work understandable and reproducible for universities, museums, artists, and cultural organizations.
What Success Looks Like
In your first 3 months, you will have shipped working prototypes for multimodal dataset ingestion, embedding generation, visualization, and exploratory search. You will have helped integrate at least one new embedding model into Gooey.AI and created a simple evaluation workflow to compare it against existing models.
In your first 6 months, you will have helped turn the best prototypes into reusable Gooey.AI workflows and core platform services. Artists, researchers, or students should be able to use these tools to upload datasets, discover meaningful relationships, interpret the results, and share their workflows with others.
Over time, your work will help Gooey.AI become a leading platform for cultural, artistic, humanities, and global impact organizations that want to explore AI in ways that are transparent, measurable, collaborative, and non-extractive.
Why This Role Matters
Most AI tools are built for search, automation, or content generation. We are interested in something broader: AI as a medium for discovery, interpretation, and shared cultural reasoning.
Latent space should not only be something engineers compute. It should become something artists, researchers, students, and communities can explore, question, critique, and reshape.
This role is about making that possible.
How to Apply
Write to jobs@gooey.ai and include a video introducing yourself and your professional superpowers and link to your portfolio/github profile.
Candidates from India, Europe/UK, and Africa preferred.
$3300 USD / month
Requirements
What We’re Looking For
You should be a strong software engineer who enjoys applied AI, creative technology, and fast prototyping.
You do not need to be an academic researcher, but you should be curious about how artists, humanities researchers, archivists, and cultural organizations work with meaning, ambiguity, interpretation, and discovery.
Required Skills
Strong Python and JavaScript/TypeScript skills.
Experience working with AI APIs, embeddings, vector search, RAG, or model orchestration.
Comfort working with APIs, queues, databases, storage, background jobs, and cloud services.
Ability to build fast prototypes and then harden the useful parts into maintainable services.
Clear communication and documentation habits.
Interest in open-source, reproducible workflows, and transparent AI systems.
Ability to work with ambiguous creative and research questions without losing engineering discipline.
Helpful Experience
Multimodal embeddings, CLIP, ImageBind, Jina, Gemini, OpenAI, or similar models.
Vector databases or search systems such as Postgres/pgvector, Pinecone, Weaviate, Qdrant, OpenSearch, or similar.
Dimensionality reduction and clustering methods such as UMAP, t-SNE, PCA, HDBSCAN, k-means, or graph-based clustering.
Data visualization libraries such as deck.gl, Three.js, D3, PixPlot-style viewers, Observable, or WebGL-based interfaces.
Generative image, audio, video, or music models.
Evaluation frameworks for LLMs, embeddings, retrieval, multilingual AI, or human-in-the-loop review.
Creative coding, digital humanities, computational art, archives, museums, cultural heritage, or research software.
Open-source contribution or experience maintaining public developer tools.
About the Company
Gooey.AI is a low-code AI orchestration platform for global impact. We help organizations build, evaluate, deploy, and share AI workflows that combine prompts, models, retrieval, tools, code, analytics, and human-readable “recipes.”
Our platform is used by global impact organizations, universities, researchers, cultural institutions, and foundations working across agriculture, health, education, arts, and public services. Our clients and partners include organizations such as the Gates Foundation, Rockefeller Foundation, Wellcome Trust, City of Seattle, Goethe-Institut, British Council, and UN-linked initiatives.
Gooey.AI is model-agnostic, open-source, and built around transparency, measurability, and sovereignty. We support private and open-source AI models, 1,500+ languages, web and messaging deployments, evaluation dashboards, reusable workflow recipes, and the ability to hot-swap models as they become faster, cheaper, smarter, cleaner, or more locally appropriate.
We are now expanding our work in artistic, cultural, archival, and humanities research. This role will help build the next generation of Gooey.AI tools for multimodal latent-space exploration: systems that let artists, researchers, students, and cultural organizations explore relationships across images, text, sound, video, archives, patterns, poetry, scientific material, and other forms of knowledge.