Use Case Intelligence RSS Search use cases Search
MongoDB Commercial and Professional Services San Francisco Bay Area

Mercor

Enterprise Knowledge Search

Mercor needed to replace a join-heavy recruiting architecture that made profiles slow to load and delayed newly interviewed candidates from appearing in search results for hours.

Read the original story Explore all use cases
Mercor reached a 100% fill rate on time-sensitive client requests as a direct result of this change.
Mercor reduced profile latency to milliseconds and improved internal search quality threefold.
Mercor’s fragmented, join-heavy architecture produced 13-second profile load times and a 6-hour lag before new candidates became searchable.
The underlying database has grown from 100 GB to 2 TB as of May 2026.

Reported impact

Business outcome

Mercor cut profile latency from about 13 seconds to milliseconds, improved internal search quality threefold, scaled its database from 100 GB to 2 TB, and reached faster fill rates on time-sensitive hiring requests.

  • Productivity Improvement
  • Revenue Growth
  • Scalability

Solution

Mercor unified its AI-native talent matching stack on MongoDB Atlas by collapsing profile data, semantic retrieval, and exact filtering into one search pipeline. Candidate information from resumes, professional profiles, social accounts, and AI-scored video interviews is denormalized into Atlas collections instead of being spread across 21 relational tables. MongoDB Vector Search on Atlas works with Voyage AI embeddings to find semantic matches between roles and candidates, while hard filters for criteria such as location or credential type are applied in the same query. Lucene-powered text search and internal ranking models remove false positives before results reach customers. Fireworks AI hosts the language models that power scoring and profile synthesis, and Apache Kafka drives change-data-capture updates so candidates become searchable as soon as new events arrive. The result is a single, scalable hiring data layer that supports rapid experimentation without operational sprawl.

Data flow

Resumes, profiles, social data, and AI-scored interview outputs are denormalized into MongoDB Atlas, embedded with Voyage AI, indexed by MongoDB Vector Search on Atlas, filtered and reranked in one query, and refreshed through Apache Kafka events so newly updated candidates are immediately available to Mercor's clients.

Solution architecture

5 components · 4 layers
  1. Ingest
    • Apache Kafka Feeds event-driven updates into the search pipeline so candidate changes become searchable immediately.
  2. Compute
    • Voyage AI Generates the embeddings Mercor uses for specialized matching tasks.
    • Fireworks AI Hosts the language models used for candidate scoring and profile synthesis.
  3. Storage
    • MongoDB Atlas Stores Mercor's denormalized candidate and job data as the unified operational data layer for search and ranking.
  4. Serving
    • MongoDB Vector Search on Atlas Executes semantic candidate-to-role matching while supporting exact filters in the same query.

Architecture clues

  • Atlas clusters scaled from M30 to M200 with auto-scaling to absorb traffic spikes.
  • Hard filters such as location are applied in the same query as vector similarity.
  • Mercor replaced joins across as many as 21 tables with a denormalized collection model.
  • The search pipeline combines vector search, Lucene-powered full-text search, and ranking algorithms.

Evidence from the source

Candidates are now searchable the moment they’re ready, rather than hours after the hiring window has opened.
“If you’re getting a profile in milliseconds, you’ll be able to search more, you’ll be able to retrieve more, and you’ll be able to hire faster.”
“In the hiring space, location should be a hard filter. It should not be a signal.”
“We don’t have to manage two separate services, and we also got early access to Voyage AI releases through MongoDB.”
Transparency Evidence & methodology 100/100 reference value · 89% confidence · 1 source

Review how this reference was scored and trace each normalized claim back to its documented source.

01

Scoring methodology

How scores are calculated

Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.

Customer reference value100/100
  • Problem clarity20/20
    Measures whether the customer's operational problem is specific enough to compare.
  • Implementation depth25/25
    Rewards named components, data flow, integrations, and architecture detail.
  • Outcome specificity20/20
    Measures whether the reported benefit is concrete rather than a generic claim.
  • Quantified impact25/25
    2 metrics describe measurable impact.
  • Deployment scale10/10
    Rewards evidence of a live or scaled deployment and meaningful integration scope.
Evidence confidence89%
  • Source authority20/25
    The main evidence is an official vendor-published customer story.
  • Claim traceability21/25
    10 of 12 key fields link to a named source.
  • Direct evidence20/20
    4 direct source quotes retained.
  • Date provenance10/10
    The publication date comes from the source page or publisher.
  • Source transparency8/10
    1 source reference document publisher, URL, authority, retrieval, and supported fields.
  • Implementation specificity10/10
    Rewards source-backed components, data flow, quantified impact, and a specific outcome.
02

Traceability

Sources & provenance

Field coverage shows which facts each source supports. Dates are shown only when documented by the source.

  1. MongoDB

    Mercor customer story

    First-party
    https://www.mongodb.com/solutions/customer-case-studies/mercor
    Source type
    Vendor-published customer story
    Published
    Jul 8, 2026
    Supports
    • Customer
    • Business problem
    • Solution
    • Products
    • Solution components
    • Data flow
    • Integrations
    • technical area
    • use case category
    • Business outcome
    • outcome category
    • Metrics
    • Architecture
    • evidence quotes
    • Publication date

Date evidence

  • Published on source page Jul 8, 2026 · MongoDB · First-party
    Open date source
Not separately sourced Industry, Region, Company size. Treat these normalized attributes as lower-confidence until a dedicated source is added.

Related

Similar use cases

See all like this
MongoDB Aug 31, 2026 LexisNexis Risk Solutions Commercial and Professional Services global LexisNexis needed an AI retrieval architecture that could preserve relationships and time context across fast-changing financial and commodities data, because standard RAG could not produce reliable answers for… LexisNexis publishes 2,000 articles daily and tracks over 2,200 global commodities. Enterprise Knowledge SearchRisk Management MongoDB Atlas MongoDB Archived · Sep 23, 2025 ICIS Commercial and Professional Services Global ICIS’s role as a source of commodities intelligence makes it a vital cog as the global economy becomes ever hungrier for information. Enterprise Knowledge Search MongoDB Atlas MongoDB Archived · Sep 30, 2024 IGT Solutions Commercial and Professional Services India IGT Solutions needed to replace manual, error-prone workflows and a split relational-plus-vector architecture with a flexible platform that could store large volumes of operational and vector data, support multiple AI… IGT Solutions operates 30 delivery centers and provides end-to-end platforms for over 25,000 users. Document ProcessingEnterprise Knowledge Search Azure · MongoDB Atlas · MongoDB Atlas Vector Search