Skip to content
HickmaTech invite-only platform
Segment. Pattern. Intersection.

MLAtlas for teams that need more than a global AUC score.

MLAtlas by HickmaTech is a governance-ready command center for post-training audits on classification models and scored datasets. MLAtlas maps score behavior across Segments, Patterns, and Intersections to reveal robustness and reliability risks that global metrics can miss.

Active audit modes
Score-driven
Run oversight
Local job + report status
Governance visibility
MLAtlas Segment Diagnostics, MLAtlas Pattern Diagnostics, MLAtlas Intersection Diagnostics
Commercial platform
HickmaTech review queue + API access
Command center preview
Portfolio evidence review
Running
Latest stage
MLAtlas Pattern Diagnostics
Pattern validity checks run before segment, pattern, and intersection robustness findings are finalized.
Risk posture
2 high-risk slices
Concentrated inside one Pattern Group and one deeper Intersection slice.
Mode
Prediction-score audit
Outputs
11 artifacts
Evidence pack
MLAtlas Evidence Pack
Problem

Serious model review needs a command surface, not scattered exports.

Audit evidence should come from one live workflow, not scattered scripts and exports.

MLAtlas turns one score-driven audit submission into a tracked organization-scoped run, generated summaries, and downloadable v2 evidence artifacts.

Reviewers need robustness, pattern, and intersection findings in one place.

MLAtlas Segment Diagnostics, MLAtlas Pattern Diagnostics, and MLAtlas Intersection Diagnostics keep global metrics, segment findings, pattern groups, and concentrated intersection signals connected to one audit record.

Teams need careful governance language, especially when higher-risk modes are gated.

MLAtlas maps score behavior across Segments, Patterns, and Intersections to reveal robustness and reliability risks that global metrics can miss.

How it works

A short workflow that still produces governance-grade outputs.

01

Upload a scored dataset

Run score-driven audits with a CSV or CSV.GZ dataset, a binary target, and a numeric score column in [0, 1].

02

Control the evaluation scope

Choose target, score, optional ID, excluded columns, optional protected columns, and safe v2 audit parameters.

03

Review live evidence outputs

Track the queue, inspect the generated findings, and download the final artifacts from the completed run.

Audit modes

MLAtlas Score Audit Engine runs two active audit paths with one gated higher-risk mode.

Mode 1: Score-driven audit

Upload a scored dataset with a score column, map the binary target, and control the v2 audit parameters directly.

Mode 2: Model-driven audit placeholder

Future sandboxed model execution can generate a score and then feed the same v2 audit engine.

Pretrained-model audit remains blocked in local and staging environments until isolated execution is enabled.
Evidence outputs

MLAtlas Evidence Pack for technical and policy review.

audit_summary.json
global_metrics.json
segment_table_v2.csv
pattern_table_v2.csv
intersection_table_v2.csv
score_error_heterogeneity.csv
operating_point_sensitivity.csv
model_card.md
MLAtlas Evidence Pack
Security and privacy

Trustworthy by default, careful about higher-risk workflows.

Score-driven audits are active in this build and do not train the audited model.
Model-driven execution remains blocked until isolated execution is enabled.
Reports are governance-ready evidence outputs, not certifications or regulatory approvals.
Pricing preview

Built as a paid command center, not a hobby audit tool.

TrialFree for 14 days
Starter$299/month
Pro$1,499/month
CustomContact us
Final call to action

Start with one reviewed pilot workspace, then scale into deeper governance controls.

Early users request access first. After manual approval, HickmaTech enables account creation so the team can sign in, start an audit, review findings, and export the MLAtlas evidence package.

Pricing preview

Trial, Starter, Pro, and Custom paths sized by usage and review needs.

Trial is for limited evaluation, Starter is for small teams, Pro is the main validation plan, and larger or regulated needs route to Custom.

Trial

Free for 14 days

Evaluation-only access for testing the workflow with limited synthetic or local data.

Row limit5,000 rows
Audit allowance3 total
File limit25 MB
API accessNo
SupportDocumentation only
Trial is a limited evaluation tier
API access and protected-attribute audit are not available
Start free trial

Starter

$299/month
$2,990/year

First paid plan for small ML or data teams running bounded validation workflows.

Row limit50,000 rows
Audit allowance10/month
File limit100 MB
API accessLimited, 1 key and 1,000 calls/month
SupportEmail, best effort
Starter is for adoption and small teams
Protected-attribute audit requires Pro or Custom
Start Starter

Pro

Most popular
$1,499/month
$14,990/year

Main validation and analytics plan for recurring model-evidence review.

Row limit250,000 rows
Audit allowance75/month
File limit500 MB
API accessYes, 5 keys and 25,000 calls/month
SupportPriority email
Pro is the main self-serve revenue plan
Private deployment, SSO/SAML, and custom retention route to Custom
Start Pro

Custom

Contact-led
Contact us

For regulated teams, larger datasets, higher-volume API users, security requirements, and private/VPC deployment.

Row limitCustom
Audit allowanceCustom
File limitCustom
API accessCustom
SupportCustom SLA
Custom is contact-sales only and never self-serve checkout
Limits are manually configured for the organization
Contact us