In our experience most companies that look for employees on social media just dully post a link to a job post on external site. We believe there is a better way.
What it does
Jobbot is a chatbot generator for smart recruitment. Each post is a separate AWS LEX bot. Each twit has a direct message button that routed messages to specified bot. There is also an option to speak to the bot directly and ask him to search for jobs.
How I built it
We used Lex Model Building Service to achieve high level of flexibility. AWS Lambda is used to create Twitter's REST API module that handle webhooks. DynamoDB is used to store data while Cognito handles user login and registration. Web page is done with JQuery. It is worth noting that each job post is a separate LEX Bot.
Challenges I ran into
Biggest challenge that we've ran into was generating lambda functions for Lex on the fly. It was really tricky and pretty hard to implement. Creating bots on the fly also require to archiving naming and enumeration system to avoid collisions and limitations of Bot and intent Naming.
Accomplishments that I'm proud of
The best part of Jobbot is flexible generator empowered by custom Twitter integration. It allows to create bots on the fly without collisions, edit them on the fly and navigate through bots by the hub. Every aspect of this process is automated including OAuth 1.0 implementation and posting advertisements. Using Twitter's DM link it is possible to chat with many bots bound to single account without troubles.
What I learned
How to use Lex in custom integrations.
What's next for JobBot
Smarter question generation. We hope that collecting questions entered by human users bot can learn to ask/generate better questions. Facebook integration, CV creation and better job search.