To generate a screening summary for each applicant
Suggested interview questions per applicant, tied to the job posting
Hiring decisions made by AI, with employers making every call
WorkOnward
New York, NY
HR Technology / Hiring Marketplace
Amazon Bedrock, Amazon EventBridge, Amazon S3, Amazon RDS (PostgreSQL)
WorkOnward is a hyper-local hiring marketplace connecting small businesses with local, deskless job seekers. Employers on the platform faced a time-consuming manual screening process after candidates applied to their job postings. Avahi delivered an AI-powered interview screening solution built on Amazon Bedrock that automatically screens applicants against job requirements, generates employer-facing summaries with suggested interview questions, and coordinates interview scheduling. The solution reduces manual screening burden for small business employers while maintaining consistency with WorkOnward’s existing candidate-side experience.
WorkOnward operates a hyper-local hiring marketplace designed specifically for Main Street employers such as restaurants and other small businesses seeking to connect with local, largely deskless job seekers. The platform handles the full hiring workflow from employer job posting through candidate matching and interview coordination. With a small engineering team balancing multiple concurrent product priorities, WorkOnward sought to extend AI capabilities to the employer side of their marketplace to reduce the manual effort required during the applicant screening phase.
After employers posted jobs on the WorkOnward platform and candidates applied through the existing matching functionality, employers faced a significant manual screening burden. Each applicant needed to be individually reviewed against job requirements, relevant interview questions had to be prepared, and interview scheduling required back-and-forth coordination. For small business owners already stretched thin, this screening process consumed valuable time that could be spent running their businesses.
WorkOnward had already built AI-assisted scheduling capabilities on the candidate side of their platform, but the employer experience lacked equivalent automation. Without addressing this gap, employers would continue spending excessive time on repetitive screening tasks, potentially slowing their hiring process and reducing the platform’s value proposition for the small businesses it serves.
Additionally, AI-assisted hiring workflows carry fairness and legal considerations, including automated employment decision tool regulations such as New York City’s Local Law 144. Any solution needed to produce screening summaries and suggestions rather than automated hiring decisions, maintaining human oversight throughout the process.
Amazon Bedrock provided the foundation model hosting capabilities WorkOnward needed for intelligent applicant screening without requiring the company to build and maintain its own AI infrastructure. Bedrock’s latency-optimized inference mode enabled screening summaries to be generated in under 10 seconds per applicant, meeting the responsiveness requirements for an employer-facing workflow. The service’s integration with WorkOnward’s existing AWS environment (PostgreSQL and Amazon S3 data stores) simplified the data flow architecture, while Amazon EventBridge provided the event orchestration needed for scheduling coordination.
This engagement followed a prior AI assessment during which Avahi and WorkOnward jointly reviewed the platform’s architecture and ranked priority AI use cases. That collaborative assessment established Avahi’s understanding of WorkOnward’s technical environment and business objectives. When WorkOnward decided to prioritize employer-side screening ahead of the originally recommended sequencing, Avahi was positioned to adapt quickly, having already mapped the platform’s data structures and integration points. The existing AWS account and IAM access arrangement between the two organizations further streamlined the engagement.
Avahi delivered an AI-powered screening solution that integrates with WorkOnward’s existing job posting and applicant data infrastructure. The solution reads structured applicant and job posting data from WorkOnward’s PostgreSQL database and Amazon S3 storage, then uses a foundation model hosted on Amazon Bedrock to evaluate each applicant against the job posting’s stated requirements.
The screening component produces a structured summary for each applicant that compares their profile against job requirements such as required skills, availability, and location. These summaries are grounded in the specific job posting and applicant data provided, with structured logging of the screening rationale to support any future bias audit or compliance review WorkOnward may need to commission.
The interview question generation component uses the same Bedrock-hosted model to produce up to five suggested interview questions per applicant. These questions are tailored to the specific job posting and grounded in the applicant’s profile. The system returns fewer than five questions only when insufficient grounded content is available, rather than fabricating content.
For interview scheduling, Avahi built a sandboxed scheduling component orchestrated via Amazon EventBridge that mirrors the interaction pattern already established on WorkOnward’s candidate-facing flow. Employers can propose an interview time, and candidates can accept or propose an alternate time. This component demonstrates the scheduling exchange in a test environment without modifying WorkOnward’s live candidate-side system, allowing the company to evaluate the pattern before committing to production integration.
Screening summaries, suggested interview questions, and scheduling prompts are delivered through WorkOnward’s existing employer notification channels rather than requiring a new user interface, enabling faster deployment and maintaining consistency with the platform’s current employer experience.
Technical architecture documentation covering the screening layer, data flow, logging schema, and scheduling coordination design
Confirmed Bedrock model selection for screening and question generation components
Functional screening prototype demonstrating AI-assisted applicant screening against job posting requirements
Interview question generation component producing job-relevant, grounded interview questions per applicant
Sandboxed employer scheduling component demonstrating interview time proposal and candidate response coordination
Sample code and integration snippets for the Bedrock-based screening and scheduling components
Working end-to-end demonstration of the screening and scheduling experience
Technical summary report with findings and recommended next steps
Knowledge transfer session covering architecture and extension paths for WorkOnward’s engineering team
The solution validates that AI-assisted screening and interview coordination can meaningfully reduce the manual screening time employers currently spend on the WorkOnward platform. By automating the comparison of applicant profiles against job requirements and generating relevant interview questions, the system shifts employer effort from repetitive evaluation tasks to higher-value candidate conversations.
The structured logging of screening summaries and their underlying rationale positions WorkOnward to meet potential compliance requirements, including the ability to support independent bias audits before any production rollout with real candidates. The sandboxed scheduling component establishes a foundation for extending the candidate-side scheduling experience to employers, maintaining platform consistency while demonstrating the technical feasibility of the integration pattern.
WorkOnward
New York, NY
HR Technology / Hiring Marketplace
Amazon Bedrock, Amazon EventBridge, Amazon S3, Amazon RDS (PostgreSQL)
WorkOnward is a hyper-local hiring marketplace connecting small businesses with local, deskless job seekers. Employers on the platform faced a time-consuming manual screening process after candidates applied to their job postings. Avahi delivered an AI-powered interview screening solution built on Amazon Bedrock that automatically screens applicants against job requirements, generates employer-facing summaries with suggested interview questions, and coordinates interview scheduling. The solution reduces manual screening burden for small business employers while maintaining consistency with WorkOnward’s existing candidate-side experience.
WorkOnward operates a hyper-local hiring marketplace designed specifically for Main Street employers such as restaurants and other small businesses seeking to connect with local, largely deskless job seekers. The platform handles the full hiring workflow from employer job posting through candidate matching and interview coordination. With a small engineering team balancing multiple concurrent product priorities, WorkOnward sought to extend AI capabilities to the employer side of their marketplace to reduce the manual effort required during the applicant screening phase.
After employers posted jobs on the WorkOnward platform and candidates applied through the existing matching functionality, employers faced a significant manual screening burden. Each applicant needed to be individually reviewed against job requirements, relevant interview questions had to be prepared, and interview scheduling required back-and-forth coordination. For small business owners already stretched thin, this screening process consumed valuable time that could be spent running their businesses.
WorkOnward had already built AI-assisted scheduling capabilities on the candidate side of their platform, but the employer experience lacked equivalent automation. Without addressing this gap, employers would continue spending excessive time on repetitive screening tasks, potentially slowing their hiring process and reducing the platform’s value proposition for the small businesses it serves.
Additionally, AI-assisted hiring workflows carry fairness and legal considerations, including automated employment decision tool regulations such as New York City’s Local Law 144. Any solution needed to produce screening summaries and suggestions rather than automated hiring decisions, maintaining human oversight throughout the process.
Amazon Bedrock provided the foundation model hosting capabilities WorkOnward needed for intelligent applicant screening without requiring the company to build and maintain its own AI infrastructure. Bedrock’s latency-optimized inference mode enabled screening summaries to be generated in under 10 seconds per applicant, meeting the responsiveness requirements for an employer-facing workflow. The service’s integration with WorkOnward’s existing AWS environment (PostgreSQL and Amazon S3 data stores) simplified the data flow architecture, while Amazon EventBridge provided the event orchestration needed for scheduling coordination.
This engagement followed a prior AI assessment during which Avahi and WorkOnward jointly reviewed the platform’s architecture and ranked priority AI use cases. That collaborative assessment established Avahi’s understanding of WorkOnward’s technical environment and business objectives. When WorkOnward decided to prioritize employer-side screening ahead of the originally recommended sequencing, Avahi was positioned to adapt quickly, having already mapped the platform’s data structures and integration points. The existing AWS account and IAM access arrangement between the two organizations further streamlined the engagement.
Avahi delivered an AI-powered screening solution that integrates with WorkOnward’s existing job posting and applicant data infrastructure. The solution reads structured applicant and job posting data from WorkOnward’s PostgreSQL database and Amazon S3 storage, then uses a foundation model hosted on Amazon Bedrock to evaluate each applicant against the job posting’s stated requirements.
The screening component produces a structured summary for each applicant that compares their profile against job requirements such as required skills, availability, and location. These summaries are grounded in the specific job posting and applicant data provided, with structured logging of the screening rationale to support any future bias audit or compliance review WorkOnward may need to commission.
The interview question generation component uses the same Bedrock-hosted model to produce up to five suggested interview questions per applicant. These questions are tailored to the specific job posting and grounded in the applicant’s profile. The system returns fewer than five questions only when insufficient grounded content is available, rather than fabricating content.
For interview scheduling, Avahi built a sandboxed scheduling component orchestrated via Amazon EventBridge that mirrors the interaction pattern already established on WorkOnward’s candidate-facing flow. Employers can propose an interview time, and candidates can accept or propose an alternate time. This component demonstrates the scheduling exchange in a test environment without modifying WorkOnward’s live candidate-side system, allowing the company to evaluate the pattern before committing to production integration.
Screening summaries, suggested interview questions, and scheduling prompts are delivered through WorkOnward’s existing employer notification channels rather than requiring a new user interface, enabling faster deployment and maintaining consistency with the platform’s current employer experience.
Technical architecture documentation covering the screening layer, data flow, logging schema, and scheduling coordination design
Confirmed Bedrock model selection for screening and question generation components
Functional screening prototype demonstrating AI-assisted applicant screening against job posting requirements
Interview question generation component producing job-relevant, grounded interview questions per applicant
Sandboxed employer scheduling component demonstrating interview time proposal and candidate response coordination
Sample code and integration snippets for the Bedrock-based screening and scheduling components
Working end-to-end demonstration of the screening and scheduling experience
Technical summary report with findings and recommended next steps
Knowledge transfer session covering architecture and extension paths for WorkOnward’s engineering team
The solution validates that AI-assisted screening and interview coordination can meaningfully reduce the manual screening time employers currently spend on the WorkOnward platform. By automating the comparison of applicant profiles against job requirements and generating relevant interview questions, the system shifts employer effort from repetitive evaluation tasks to higher-value candidate conversations.
The structured logging of screening summaries and their underlying rationale positions WorkOnward to meet potential compliance requirements, including the ability to support independent bias audits before any production rollout with real candidates. The sandboxed scheduling component establishes a foundation for extending the candidate-side scheduling experience to employers, maintaining platform consistency while demonstrating the technical feasibility of the integration pattern.
Let’s explore your high-impact AI opportunities together in a complimentary session
