Fynd Engineering X
Cutting-edge research on productivity enhancement throughout the software development life cycle (SDLC). We harness the potential of AI to help teams design better, streamline production and deploy projects quicker than ever before
Our Objectives
To create an exhaustive repository of all code used across Fynd’s platforms. This would serve as a developer portal and a training dataset for future LLM models
To accelerate and streamline software development processes using generative AI
To reduce the amount of time needed to design, develop and deploy software applications
To create a one-of-its-kind LLM model for
Who can benefit from our research?
Development teams
Software development teams looking to accelerate and streamline their entire development cycles
Coding enthusiasts
Individual developers seeking help/guidance for their projects
Tech businesses
IT and tech-related businesses aiming to establish quick, standardised development processes within their organization
Focus areas & current projects
Fynd Code Wikipedia
An exhaustive repository of all code used across Fynd’s platforms. This would serve as a comprehensive and accessible developer portal helping developers streamline their work in the Fynd ecosystem
The Fynd Code Wikipedia will serve as the basis for
- All code optimisation, summarisation, testing and scaffolding for Fynd’s internal projects
- Code auto-completion: SFT integration on code editors like VScode as an Extension
- Fynd Extension generator
- Fynd Theme generator
- An SRE-GPT
What's next
- Generating instructions for all of Fynd’s code. Once the data preparation is completed, supervised fine tuning will follow.
- From the above use cases, we are currently working on Fynd Extension generator & code auto-completion
TigerBoost Coder
A suite of LLMs with knowledge of Fynd’s codebase. It will generate code on demand and boost development productivity. TigerBoost LLM coder will generate multiple drafts of code using multiple LLMs and provide the option to the users to select best suited code
What's next
- Create embeddings of all Fynd code
- Create Vector database to store embeddings
- Extract relevant data and augment prompts
- Add OpenAI APIs to generate code augmented with context
Fynd Themes & Extension Generator
A fine-tuned LLM model to help developers rapidly generate high-quality and functional code for a wide range of themes and extensions, reducing the time and effort required in manual coding. It would also be capable of converting a given website to a comprehensive and accurate Figma file
The process for generating and using a theme/extension on the Fynd Commerce Platform would be as follows:
- Select a third-party website to clone and create a Fynd Commerce Platform theme
- Convert website to figma design
- Use figma design and GAI to create Fynd Platform theme components
What's next
Exploring a website-to-figma-to-code approach, generating a customisable HTML file
Playground
Try the latest innovations for yourself
Collaborate with us
We are open to research collaborations to further enhance our solutionsContribute to or utilize our open-source projects
Work with us
Want to push the boundaries of tech innovation with us? Join our team!
