Founding Software Engineer
We believe that too much human brainpower is wasted on repetitive tasks; and that individuals should focus on creativity, decision-making, and communication — while AI handles the tedious execution.
That's why we started Twin. We aim to create the productivity platform to enable SMB teams to automate complex and repetitive tasks across multiple apps - as seamlessly as delegating them to smart interns.
Twin Labs was founded on 09/01/2023, after Joao & Hugo released a working prototype of Twin core technology: a headless browser coupled with vision models able to autonomously control graphic user interfaces.
We are now growing our founding team to:
- Advance our research
We’re working on a new architecture for AI agents, which is more reliable and extensible than the state of the art. This involves the research and implementation of novel methods for extracting accurate decisions from machine learning models, notably large language models, mixing a pragmatic approach of leveraging existing models with exploring new techniques for environment and action embeddings, augmented generation, decision pipelines, few-shots prompting, among others.
- Ship a first version of our app to selected B2B design partners
We put a high bar on delivering a reliable and seamless experience, integrating with the daily workflow our customers.
Tasks
As a founding software engineer, you will be part of a small team, including software and research engineers, implementing Twin’s core product —- a platform to automate repetitive tasks using AI agents. This platform includes a web application, a browser extension which can take actions for the user, a chat-based interface for communicating with the AI agents, and the core agent service which makes decisions given the current environment and the goal.
As an early member of a small team, software engineers at Twin get to have significant impact on the product and are exposed to the whole stack, from experimenting and testing new ideas in the field of AI agents, to implementing robust backend systems to support our core product.On a typical week, you might work on implementing a novel architecture for a module of our agents, improve our evaluation framework, fix bugs in the backend, optimize database queries, improve the performance of the extension, among other tasks.
If you love working on hard technical problems, find the field of AI agents fascinating, and have the ability to own problem end-to-end, this position is for you.
Key responsibilities- Implement novel techniques to increase the accuracy of our AI agents
- Implement new features, fix bugs, and optimize the backend and services
- Write infrastructure code, deploy and monitor services on the cloud
- Work with open-source multi-modal models, to improve agent decision pipelines
- Generate and filter data for training and finetuning custom models
- Improve pipeline performance with finetuning and prompt augmentation
- Improving performance of knowledge retrieval
- 3+ years of relevant engineering experience
- Proficiency in Python and Typescript
- Familiarity with large language models