Computer Vision Engineer (OCR)

About Farsight Vision

Farsight Vision develops AI-powered geospatial intelligence solutions for defence and security operations. The platform transforms raw aerial and sensor data into real-time 3D environments, actionable analytics, and AI-assisted operational scenarios, enabling enhanced situational awareness, mission planning, and faster decision-making for both human operators and autonomous systems. We operate at the intersection of geospatial data, computer vision, and defense operations — and if that interests you, read on.

We are looking for a Senior Computer Vision Engineer to own text recognition within our platform end-to-end. That means turning maps, documents, imagery, and video into structured, reliable data, from the first dataset to a model running in production. This is not a role for someone who waits for a fully scoped ticket. You will define the problem, choose the approach, and stay accountable for how it performs in real operational use.

Responsibilities

• Design, build, and deploy production OCR pipelines covering text detection, recognition, and post-processing for images and video in GIS and defense contexts.

• Own recognition quality on hard real-world input: low resolution, motion blur, perspective distortion, noise, partial occlusion, and mixed or handwritten text.

• Build and maintain multilingual recognition across Ukrainian, Russian, and English, including mixed-script and domain-specific vocabulary.

• Develop layout analysis and structured extraction, turning maps, tables, forms, and scanned documents into usable data rather than raw strings.

• Create training data at scale through annotation workflows, augmentation, and synthetic text generation, and close the loop with error analysis on production data.

• Define evaluation methodology (CER/WER, field-level accuracy, latency) and use it to drive model decisions, not just to report results.

• Optimize models for speed and reliability, and integrate OCR components into operational products together with cross-functional engineering teams.

• Diagnose and fix production issues quickly, with a bias toward solving the problem over waiting for perfect inputs.

Requirements

• 5+ years of hands-on experience in computer vision and applied machine learning, with at least 2–3 years focused on OCR or scene text recognition in production.

• Expert-level Python. You write clean, efficient, production-quality code.

• Deep working knowledge of PyTorch and OpenCV, and the full model development workflow from data to deployment.

• Proven experience with modern text detection and recognition architectures, such as DBNet, CRAFT, EAST, CRNN/CTC, attention-based or transformer-based recognizers (TrOCR, Donut, PARSeq).

• Practical experience with OCR frameworks and engines, such as PaddleOCR, MMOCR, docTR, EasyOCR, or Tesseract, and a clear understanding of when to use them versus building custom models.

• Experience with document understanding and layout analysis, including structured information extraction.

• Strong grasp of data work for OCR: annotation, synthetic text data generation, augmentation, and dataset quality control.

• Ability to translate research into reliable engineering outputs with minimal supervision.

• Bachelor's or Master's degree in Computer Science, Artificial Intelligence, or a related field, or equivalent demonstrable experience.

• Upper-Intermediate+ English.

Nice to have

• Experience with Cyrillic and multilingual OCR, handwriting recognition, or text on maps and aerial imagery.

• Experience with vision-language models (VLMs) for document and image understanding.

• Model optimization and edge deployment (ONNX Runtime, TensorRT, OpenVINO; NVIDIA Jetson or similar).

• Background in GIS data, geospatial imagery, or defense-tech products.

Security Eligibility

Our platform is actively used by the Armed Forces in live operations. As an Estonian-Ukrainian defence technology company handling sensitive operational data, we have a responsibility — legal, ethical, and practical — to ensure the integrity of everyone on our team.

All candidates must be able to confirm the following:

• Neither you nor your close relatives (parents, spouse, siblings, children) reside in Russia, Belarus, or temporarily occupied Ukrainian territories.

• You have no regular personal, financial, business, or professional contact with individuals in those territories.

• You consent to a deeper security verification if the process advances to later stages.

If any of the above criteria are not met, please do not proceed with your application.

Why Farsight Vision

We are a startup operating under real-world conditions — our product is in active deployment, and the pace reflects that. The text you teach our systems to read ends up in front of people making decisions in the field, so results matter far more than process here. If you enjoy owning hard problems, learning new tools fast, and finding solutions instead of reasons why something can't be done, you'll fit in.

Flexible working hours. Paid vacation and sick leave. Direct access to founders and leadership.

Application Process

1. Apply with your CV and a brief cover letter explaining your interest and relevant experience.

2. Complete a technical assignment built around a real OCR problem we solve.

3. Interview with our technical team and leadership.