How Hypefactors Turned Losing Data in Slack Into Smooth Collab in Neptune
Hypefactors is a media intelligence company that utilizes machine learning to automate Public Relations workflows and analyze brand reputation across various media, including social media, print, TV, and radio.
The challenge
To analyze every form of data, including images, text, and tabular data, Hypefactors work on a variety of ML problems, from NLP classification to computer vision segmentation to regression for business metrics.
As they train and improve many enrichment models using different ML techniques, this naturally involves running many experiments and articulating ways to store the metadata generated by those experiments.
Initially, the team’s experiment tracking was manageable with informal systems like Slack and personal notes; however, as the number and complexity of experiments grew—driven by an increase in models, features, and team size—these methods became inadequate.
This led to a bottleneck in their workflow, severely impacting their ability to efficiently manage and share experiment outcomes and metadata.
Standardization of tracking methods
The sudden burst in experiments meant hundreds of variations of datasets, model architectures, and corresponding outcomes. Each team member had their own system for storing and organizing this data. The lack of standardization in tracking experiments and metadata led to inconsistencies, errors, and difficulties in comparing results.
Neptune addressed these challenges by providing a centralized platform where all metadata and model artifacts could be uniformly stored and accessed. Every experiment’s metrics and outcomes were now consistently logged, making comparisons straightforward and reliable. The uniformity also reduced the time spent on managing data, allowing the team to focus more on analysis and less on administrative tasks.
Improving collaboration by replacing Slack with Neptune
As the number and complexity of experiments at Hypefactors escalated, using Slack to share and discuss experiment results became impractical. The platform was not suited to handle the complex and high-scale data involved in ML experiments.
This led to team members losing track of important experiment details, such as checkpoints or other metadata, as there was no effective way to store large volumes of information on Slack. It also generated confusion around who does what. Multiple people were doing experiments on the same problem because they could not effectively communicate.
Neptune completely changed how team members collaborated on projects. By giving them one single source of truth and enabling the sharing of URL links to experiments, Neptune facilitated easy access to detailed experiment results for all team members.
This feature significantly streamlined communication, as team members could directly view and discuss the results through Neptune’s interactive dashboards.
The results
- Streamlined access and organization of experiments and metadata;
- Improved and simplified team collaboration;
- Enhanced efficiency in managing and comparing experiments, resulting in accelerated decision-making.
Thanks to Viet Yen Nguyen and Andrea Duque for their help in creating this case study!
Product resource 
Related article 
