About Customer:
Customer is a Noida-based, tech-driven marketplace founded in 2020 that connects contractors and companies with India’s construction and daily-wage workforce. The platform helps workers build digital profiles capturing skills, experience, availability and wage expectations, and gives employers verified access to blue-collar talent through dedicated mobile apps for both sides of the marketplace – reducing dependence on physical ‘labour chowks’ and intermediaries.
Industry:
Workforce Marketplace/HR-Tech
Offering:
AWS Bedrock + GenAI (Claude Sonnet 4.5 + Nova Pro)
Business Challenges:
- Job-to-worker matching relied on rigid, static filters, missing real-world fit and slowing time-to-hire.
- Manual FAQ creation per job post couldn’t scale across volume or multiple Indian languages.
- Phone-based hiring calls happened off-platform, so payment discussions and fraud went undetected.
- Growing call volume made manual policy-violation review impractical to sustain.
- Repeated recommendation queries risked driving up inference cost and latency.
Solution Implemented:
- A Bedrock (Claude Sonnet 4.5) recommendation engine that turns intent + schema context into constrained SQL, ranking candidates via Amazon RDS.
- A daily EventBridge/Lambda pipeline using Nova Pro to generate structured, multilingual FAQs per job post, written back to RDS.
- A Step Functions pipeline where Transcribe converts calls to text, Bedrock flags risk signals, and Comprehend extracts sentiment and PII.
- An S3 + CloudFront analyst dashboard, fed by DynamoDB risk scores, so teams review only high-risk calls.
- Amazon MemoryDB for Valkey as a semantic cache, serving repeated queries without re-invoking Bedrock.
Services Used:
- Amazon Bedrock (Claude Sonnet 4.5, Nova Pro)
- AWS Lambda
- Amazon API Gateway
- Amazon S3
- Amazon DynamoDB
- Amazon MemoryDB for Valkey
- AWS Step Functions
- Amazon Transcribe
- Amazon Comprehend
- Amazon CloudFront
- AWS Secrets Manager
- Amazon CloudWatch + AWS CloudTrail
- Customer’s Existing Amazon RDS
Business Benefits:
- Higher match relevance, 8–10s latency target – LLM-ranked recommendations beat static filters, kept fast via caching.
- ~85% FAQ accuracy target – automated multilingual FAQs cut repetitive support queries.
- ~90% risk & sentiment accuracy targets – Bedrock and Comprehend flag fraud risk per call.
- Call coverage scales with volume – automated triage means analysts review only high-risk calls.
- Lower inference cost, ~US$4,850/month run cost – semantic caching cuts repeat Bedrock calls.
Ready to Power Your Workforce Marketplace with AI?
Click Here to Get a Free GenAI Readiness Assessment
Frequently Asked Questions
Rapyder helped a blue-collar workforce marketplace move from static, filter-based matching and manual FAQ writing to a unified GenAI platform on AWS Bedrock - improving match relevance, automating multilingual job-post FAQs, and flagging call-based fraud automatically.
By using Amazon Bedrock (Claude Sonnet 4.5) to turn employer and worker intent into schema-aware, constrained SQL queries against DLC's existing database, ranking candidates on skill, wage fit, experience and location rather than static filters - kept fast with RDS Proxy pooling and MemoryDB semantic caching.
A daily EventBridge-triggered Lambda pipeline uses Amazon Nova Pro to generate structured FAQs (question, answer, language, confidence) per job post, which land in a draft/review state so only approved content publishes.
Call recordings are transcribed with Amazon Transcribe, then analysed in parallel by Bedrock for risk signals like off-platform payment intent and by Amazon Comprehend for sentiment, key phrases and PII - orchestrated end to end by AWS Step Functions with built-in retry handling, so only high-risk calls reach the analyst dashboard.
Depending on the use case, organizations can benefit from:
- More relevant, faster candidate and job matching
- Reduced manual content and support workload
- Automated fraud and risk detection at scale
- Lower inference cost through semantic caching
- Serverless architecture that scales without added operations overhead
While this case study focuses on a blue-collar workforce marketplace, Rapyder's Generative AI and cloud modernization capabilities can be applied across industries such as BFSI, Healthcare, Manufacturing, Retail, Media & Entertainment, Logistics, and Enterprise SaaS.
