Job Details
- Department: Data / Artificial Intelligence
- Tools: Picsellia, CVAT, Notion (or equivalent)
- Contract Type: Permanent (CDI) – Full-time
- Required Experience: 4 to 6 years, including at least 2 years on a Computer Vision annotation pipeline
Purpose of Role:
The annotation team produces training datasets for object detection models, including bounding box annotation, segmentation, and characterization, using Picsellia and CVAT. Campaigns are jointly defined by the Product and AI teams and executed by a team of six annotators supervised by a technical specialist.
Core Responsibilities:
1. Guidelines Ownership
- Centralize and maintain all annotation guidelines (Bounding Box Annotation, Segmentation, Characterization)
- Implement version control and change management for all guidelines.
- Structure each guideline with Edge cases, visual examples, clear decision rules
- Establish a formal guideline validation process before campaign launch.
- Manage class changes (additions, deletions, and modifications) with full traceability.
- Ensure no campaign is launched with incomplete, conflicting, or unclear instructions.
- Goal: Zero campaigns launched with incomplete or conflicting rules.
2. Campaign Pipeline Management
- Maintain a visible and prioritized campaign backlog shared with Product and AI teams.
- Anticipate workload requirements with a minimum three-week planning horizon.
- Handle priority changes: absorb, arbitrate, and communicate decisions to the team
- Coordinate campaign creation in Picsellia and CVAT alongside AI teams.
- Monitor campaign progress and ensure timely delivery.
- Goal: Zero annotators waiting for tasks.
3. Structured Cross-Team Communication
- Run a weekly Product × AI × Annotation synchronization with written meeting notes.
- Create clear and actionable campaign briefs before campaign launches.
- Act as a single point of contact for annotation-related requests from Product or AI
- Escalate and resolve blockers before they impact production timelines.
4. Data Quality Management
- Define and enforce annotation quality thresholds (eg. rework rate <10%, target IoU per task type)
- Structure and manage Review → Rework workflows.
- Analyze quality issues and identify recurring error patterns.
- Implement corrective actions through (Training, Guideline updates)
- Validate datasets before handoff to Machine Learning teams.
- Goal: Maintain a team rework rate below 10%.
5. Tool Cleanup & Proper Usage
- Audit the current state of campaigns in Picsellia and CVAT (statuses, Naming, Missing data)
- Define and enforce standard workflows: Review -->Rework -->Done workflow
- Establish campaign tracking standards and reporting processes.
- Collaborate with the IT Director on Picsellia and CVAT API integrations
- Improve operational visibility through reliable real-time tracking.
- Ensure all active campaigns are properly managed within the tools.
- Day 30 Goal: 100% of active campaigns correctly statused in Picsellia and CVAT.
6. Team Management & Development
- Manage and mentor a team of six annotators
- Conduct goal setting, performance reviews, and regular feedback sessions.
- Support professional growth and skills development.
- Collaborate with the Annotation Technical Specialist on complex annotation cases.
- Design and maintain onboarding processes for new annotators.
- Foster a culture of accountability, documentation, and continuous improvement.
- Encourage proactive feedback and knowledge sharing within the team.
Required Skills:
Core Competencies:
- Computer Vision Annotation Pipeline Management
- Annotation Guideline Creation & Governance
- Cross-Functional Communication
- Picsellia and CVAT Operations
Technical Skills:
- Experience managing production Computer Vision annotation pipelines.
- Strong understanding of: Bounding Box + Segmentation pipeline in production
- Proficiency with Picsellia and/or CVAT.
- Ability to create structured annotation guidelines and documentation.
- Knowledge of quality metrics: IoU, Rework Rate, Inter-Annotator Agreement
- Understanding of Machine Learning data requirements and priorities.
- Experience working with annotation tools such as: Picsellia, CVAT, Labelbox, Scale AI, or equivalent platforms
Behavioral Skills:
- Excellent written and verbal communication skills.
- Ability to create clear, actionable, and unambiguous documentation.
- Proactive mindset with strong anticipation of operational risks.
- Strong organizational and prioritization skills.
- Ability to challenge and reject unstructured requests when necessary.
- High level of documentation discipline and process ownership.
- Calm and effective leadership style.
- Strong stakeholder management capabilities.
Candidate Profile:
- Master's Degree (Bac+4/5) in:
- Computer Science
- Data Science
- Computer Vision
- Technical Project Management
Experience:
- 4–6 years of professional experience.
- Minimum 2 years working within a production Computer Vision annotation pipeline.
- Experience collaborating across Product, AI, and Operations teams.
- Professional English proficiency for documentation and communication with international stakeholders.
- Experience in fast-paced environments requiring cross-functional alignment.
Success Metrics
First 30 Days
- 100% of active campaigns correctly statused in Picsellia and CVAT.
- Tool usage conventions documented and communicated to the team.
- First Product × AI × Annotation sync completed with meeting notes shared.
- Existing annotation guidelines audited and gaps identified.
Within 3–6 Months
- 100% of campaigns launched with complete and validated guidelines.
- Average team rework rate ≤ 10%.
- Campaign backlog planned on a rolling three-week basis.
- Zero unplanned workload gaps for annotators.
- Weekly Product × AI × Annotation sync established with documented meeting notes.
Benefits:
Work Location: Remote
