# AI Data Operations Analyst **Company:** Variene AI **Location:** Athens, Greece **Workplace:** hybrid **Employment type:** Full-time [Apply for this job](https://jobs.workable.com/view/eeb77ba9-833c-44ff-80ea-628d23e49a27) ## Description ### About Variene Variene develops advanced AI and computer vision software for defense and autonomous systems, designed to operate in demanding real-world environments. Our technology brings intelligent perception to platforms operating across ISR, autonomous systems, and other mission-critical defense applications. Our models are only as good as the data behind them. ### The Role As our AI Data Operations Analyst, you will own the full lifecycle of the data that trains our computer vision models: from finding the right video and imagery, to preparing and categorizing it, assigning it to our annotation team, checking the results, and reporting on what the data tells us. You will sit in management and product meetings to understand our current and upcoming projects, and you will translate those priorities into the right material, at the right time, for the annotation team. You will be the bridge between the people who set product direction and the people who create our training data, and you will bring the numbers and the feedback back to product development. Reports to: Head of Product / CTO. Works closely with: Leadership, ML engineers, DevOps, and the Annotation Team Manager. Good to know: due to the nature of our work, some of the material you will review may include footage from real conflict zones, which can occasionally be distressing. We take this seriously and will discuss it openly during the hiring process. ### What You'll Do - Take part in management and product meetings to stay aligned with active and upcoming projects. - Translate project goals into concrete data needs: target classes, scenarios, environments, and edge cases. - Agree with product development on priorities for the annotation team and keep them visible and up to date. - Independently search for, evaluate and secure video and imagery (including FPV, aerial, maritime and thermal footage) relevant to each project. - Judge each source for relevance, quality, volume and suitability for secure, off-cloud use. - Check, together with the DPO/Legal function, that we have the right to use each source (licensing, platform terms, personal data) before it enters our pipeline. - Maintain a rigorous, up-to-date source register. - Edit and prepare material for annotation: clipping, trimming, format conversion, frame extraction, de-duplication and cleaning. - Categorize and tag material with consistent metadata (class, scenario, sensor type, conditions, source) so every clip is easy to find and trace. - Package targeted batches and assign them to the annotation team in CVAT, with clear instructions on what to look at and what to look for. - Verify the integrity and completeness of every batch before it reaches the team. - Act as a second level of quality control: run sample-based reviews of completed annotations, in coordination with the Annotation Team Manager, who owns day-to-day review. - Check that annotations are not only correct, but useful for what the model needs (coverage, consistency, edge cases). - Collect annotator feedback and recurring edge cases and turn them into clearer guidance and better batches. - Track and analyze the data coming from our tools (CVAT and internal systems): annotation throughput and turnaround time, number of annotated objects per class, rejection and correction rates, and backlog. - Spot gaps and imbalances in the datasets (e.g., classes or scenarios that are under-represented) and recommend what to source next. - Produce clear, regular reports and dashboards for leadership and product development. - Collect productivity and quality metrics in line with Variene's internal policies and data protection requirements. - Document sources, preparation steps and dataset versions so every dataset is traceable and repeatable. - Follow Variene's security and data-handling procedures for off-cloud work. ## Requirements ### Knowledge and Experience - 2–4 years of experience in AI/computer vision data operations, data annotation (lead, QA or coordination roles), dataset curation, or a similar data-focused role. - Hands-on experience handling and editing video data (clipping, conversion, frame extraction) with tools such as FFmpeg. - Basic Python scripting for downloading, organizing and processing files and data. - Strong Excel skills (pivot tables, charts); ability to turn raw numbers into clear reports. - Proven ability to research and source material independently and assess its quality. - Understanding of how annotation works and of common annotation formats (e.g., COCO, YOLO). - Excellent written and spoken English; Greek is a plus. ### Highly Valued Experience or familiarity involving: - Annotation platforms such as CVAT, Label Studio or similar, including their analytics and reporting features. - Power BI or similar dashboarding tools. - Dataset management tools (e.g., Roboflow, FiftyOne). - Defense, maritime or aerial footage; ISR data workflows; thermal (IR) imagery. - Open-source intelligence (OSINT) research methods. - Working in on-premise or air-gapped environments. - Eligibility for a national or NATO security clearance. ### Professional Profile We are looking for someone with: - Initiative and high ownership, comfortable finding their own way to the data we need. - Exceptional data hygiene habits and an eye for detail. - Structured, analytical thinking and comfort with numbers. - Clear, diplomatic communication across leadership, engineers and annotators. - Curiosity about how data quality directly affects model performance. - Discretion and a strong sense of responsibility when handling sensitive material. ### What Success Looks Like Success is not measured by data volume alone. Within your first months, you will help Variene: - keep the annotation team supplied with relevant, well-prepared material, with no idle time; - make sure annotation effort is always aimed at what product development needs most; - build balanced datasets, with clear visibility of coverage per class and scenario; - reduce rework through better batches, clearer guidance and early quality checks; - give leadership a clear, regular picture of data progress, quality and gaps. ## Benefits - Competitive compensation based on experience. - Performance-related compensation.