Inspiration: AI that enables jobs and manufacturing flywheel of India through superior skill matching and trust based signals through professional networking.

What it does: Skilleasy AI enables skill matching through an intelligent taxonomy that dynamically maps jobs, skills, machine, employer type (small business, big corporations), etc. This ensures we exactly map the worker proficiency to enable them to connect with exactly suited role and with connections who can get them introduced to a given role.

How we built it: This is a partially functional prototype with a full public release planned in another quarter. It was built with a combination of LLMs and taxonomy services that ride on a knowledge base designed to probe workers to understand their skills. The probing works in a natural way without being intrusive or leading.

Challenges we ran into: Defining the feature set and key challenges we need to solve for grey collar hiring challenges (Networking and Skill matching). Thereafter, to enable networking and skill matching we needed to create an architecture which can give deterministic outcomes that guide a worker to network with specific individuals or to apply to a specific job based on their exact skills.

Accomplishments that we're proud of: Progress so far is encouraging and we really look forward to actual customer anecdotes, once we launch this to customers

What we learned: We learned the nuances of choosing LLMs for not just building fast but also taking into account deterministic reasoning, as well as costing and latency related customer experience factors. This is what led us to create our taxonomy service which is the backbone of entire Skilleasy initiative.

What's next for Skilleasy: Public launch

Built With

  • bedrock
  • codex
Share this project:

Updates