CVEDIA
CVEDIA
CVEDIA dezvoltă software de analiză video și viziune computerizată bazat pe inteligență artificială, folosind date sintetice pentru a crea și perfecționa sisteme de recunoaștere vizuală. Platforma CVEDIA-RT reunește analize preconfigurate, care pot fi adaptate pentru diferite aplicații și implementate pe diverse platforme hardware, inclusiv pe dispozitive edge. Abordarea urmărește să reducă dependența de colectarea și procesarea unor volume mari de înregistrări din lumea reală, susținând totodată cazuri de utilizare precum protejarea faunei sălbatice și imagistica termică. CVEDIA oferă și dezvoltare personalizată și asistență continuă pentru produse implementate în domenii precum inteligența artificială, logistica și consultanța.

Remote AI Engineer, Computer Vision

Build agent-driven computer vision models, synthetic data pipelines, and training workflows for CVEDIA’s production edge systems. Improve detectors, classifiers, and vision-language models for customers and partners.

Descriere

  • Maintain and improve synthetic data pipelines and automated training systems.
  • Own a model and improve it through an agent loop that generates training data.
  • Build automated cycles in which agents train detectors and classifiers, test them against validation sets, and refine them with minimal human input.
  • Develop systems to address field false positives, flag uncertain detections, and automatically correct other issues.
  • Lead vision-language model research, fine-tune models for specific tasks, and use agents to track current techniques.
  • Address customer and partner detection and classification problems by building or extending models.
  • Apply agents to QA automation and codebases that identify and fix their own bugs.
  • Share effective approaches with the team and present results to customers when needed.

Cerințe

  • Advanced English proficiency at C1 level, including fluent communication in complex, fast-paced technical and professional settings.
  • Demonstrable experience building personal agent harnesses or agentic workflows.
  • Experience operating many agents at once, including managing communication between agents, token usage, and verification of completed tasks.
  • Professional software engineering experience on both Windows and Linux.
  • Experience with Git and an issue-tracking system.
  • Experience fine-tuning or quantizing open-source models, such as training LoRAs.
  • Experience running open-source models locally.
  • Knowledge of model development from early GPT models to current open-source models, including their strengths and limitations.
  • Clear communication with colleagues and customers.
  • Strong interest in AI and agents, with a willingness to experiment with new tools and ideas.
  • Submit a resume or CV in English only.
  • Applicants from all backgrounds and countries are welcome.

Beneficii

  • Fully remote work.
  • Daily access to Claude, Codex, and self-hosted models, with no limits on applying them to work.
  • Work on models already running in production on edge devices around the world.
  • Work directly with the whole team and influence company-wide AI workflows.
  • Collaborate with colleagues interested in agent behaviour, harness design, and model quirks.

Locuri de muncă similare

Canals

Senior Machine Learning Engineer, Colombia Remote

Canals

Lead production machine learning for Canals’ logistics automation platform, from data pipelines through deployment and monitoring. Build customer-focused AI features and guide technical practices as the company grows.

Deschide
CircuitHub

Computer Vision Software Engineer

CircuitHub

Build computer vision systems for CircuitHub’s robotics-powered electronics manufacturing platform, including inspection, component picking, alignment, and real-time GPU vision.

Deschide
CrowdStrike

Threat Analyst, Machine Learning Detection Efficacy

CrowdStrike

Assess malware and customer detection reports to help improve CrowdStrike’s machine learning detections. Support false-positive management and share efficacy findings with internal teams.

Deschide