01
First purchase
The Pilot
The first paid engagement: a bounded test of one task and one rubric before either team commits to scale.
- Window
- 5 working days
- Pod
- 10–20 vetted trainers
- Receive
- Sample outputs + QA report
Services / configured
Build the evaluation pod your model needs.
Imbazo builds and runs managed human evaluation programs. Choose the decision, skill profile, and evidence you need back; start with a paid pilot, then continue with a dedicated pod after the pilot proves fit.
Start with a pilotStep 01 / choose the judgment
What do humans need to decide?
01 / Evaluate
Find the signal between two plausible answers
A calibrated pod compares model outputs, applies your rubric, records the reason for each judgment, and routes disagreements for review.
- Best for
- Preference ranking, SFT review, reward-model signals
- Pod
- Task-tested evaluators + a quality reviewer
- You receive
- Scored outputs, reviewer notes, disagreement patterns
Step 02 / choose how we work together
Start with proof. Continue with capacity.
Imbazo’s core service is managed evaluation capacity. The pilot is the first paid engagement: a controlled way to compare output quality and program economics before moving into recurring delivery.
02
Core service
The Dedicated Pod
Ongoing managed evaluation capacity configured around a recurring model or data workflow.
- Cadence
- Recurring delivery
- Team
- Skill-matched reviewers
- Control
- Calibration + ongoing QA
03
Extension
Specialist Review
Domain, language, coding, math, or safety judgment layered into a broader evaluation program.
- Match
- Language or domain skill
- Use
- Complex and edge-case tasks
- Receive
- Reviewed outputs + issue patterns
African-founded / operating advantage
Lower cost base. Same quality bar.
- Africa-based delivery with a lower operating cost base
- Trainer screening against the task
- Rubric calibration before full volume
- Independent review and QA sampling
- Disagreement and escalation handling
- Reviewer notes and error-pattern reporting
The first five days
Proof before ongoing delivery.
IMB / QA–01
A reviewable body of evidence
- 01Sample outputs from the calibrated task
- 02Reviewer notes showing the reason behind judgments
- 03Recurring error and disagreement patterns
- 04A QA report with scale-up recommendations
Final scope depends on task design and is agreed before the paid pilot begins.
Start a conversation / built around the work
Bring us one decision your model needs humans to make.
We will reply with the questions, team shape, and first engagement that make sense—whether that is a five-day pilot, specialist review, or an ongoing pod.
Prefer email?tina@imbazo.com