Thalanor - R+D Engineer - Autonomy and Vision

In the last 3–4 years, which option best describes your work with computer vision and deep learning algorithms in production (not just academic prototypes)? *

I have designed and trained detection/segmentation models (e.g. YOLO-type, Faster R-CNN, vision transformers) and brought them to production in real systems, tuning both the model and its integration into the product software.
I have worked continuously with vision models (detection/classification), adapting existing architectures and taking part in their integration into test systems or pilots, while others handled the final deployment.
I have mainly used pre-defined vision models (e.g. from common frameworks), tuning hyperparameters or small details, mostly in lab projects, theses or internal tests without strong production pressure.
My experience with computer vision is occasional (a project or a course), I tend to rely on “ready-made” models and my main focus has been other areas such as backend, apps, data or more theoretical research.

Thinking about your ideal daily work in R&D, which of these descriptions fits you best when you face a new guidance or control algorithm that nobody in your team has used before? *

I’m comfortable starting from optimal control/statistics papers, deriving the formulation, testing it on the whiteboard/simulation and then implementing it in a large repository, integrating it with the rest of the system until it works on the real vehicle.
I like to study the mathematical basis and experiment with the algorithm in simulation, and I usually need occasional support for full integration into the system code or for tuning more advanced control details.
I prefer someone to define the basic formulation and code structure; from there I can implement simple versions and help with tests, but I tend to focus on well-bounded parts of the system.
I feel more aligned with roles where guidance algorithms are already defined and my contribution is mainly to use existing tools or follow closed specifications, without needing deep mathematical involvement.

Which of these situations best describes your experience and comfort level integrating your R&D work into large existing codebases (tens of thousands of lines) used in production? *

I have worked inside large codebases, located relevant modules, understood how my algorithm interacts with the rest of the system and made structural changes without breaking overall stability, including tests and ongoing maintenance.
I have worked on projects with relatively large code, adding new modules or functions and adapting existing ones, usually following structural guidelines and reviews from more senior colleagues.
I usually work in small or isolated repos (e.g. research prototypes, notebooks, standalone services), and when I contribute to large projects my participation is limited to well-scoped components.
My typical R&D programming work is in independent prototypes, notebooks or academic projects, without needing to care much about integration into complex deployed systems.

Regarding work environment and professional lifestyle, which option matches you best for the next few years in an R&D role in defence/autonomous drones? *

I’m excited by working in a deep-tech defence startup, with high R&D investment, field test trips, intense periods when needed and regular on-site presence; I value ownership over cutting-edge technology even if it’s demanding.
I’m comfortable with a demanding environment and mostly on-site work, provided there is some flexibility (e.g. occasional remote days) and a clear technical purpose linked to challenging projects.
I prefer most work to be remote or in a stable environment, with predictable hours and load, and to take part in R&D without too much pressure from field tests or tight deadlines.
I feel more aligned with large, highly structured companies, with clear processes, multiple stakeholders and longer development cycles, even if the R&D part is less intense or less focused on complex physical products.

Thinking about your ideal contribution during your first year in an R&D–Computer Vision Engineer role, which option is closest to what you typically deliver? *

Deeply understand existing vision and guidance algorithms, identify why they fail in edge cases, propose and test new state-of-the-art techniques, and leave at least one clear improvement running on the drone and validated in the field.
Understand current algorithms well, help stabilize them and propose some improvements based on known techniques, achieving solid results in simulation and partially contributing to real tests.
Get familiar with the codebase and models, make incremental adjustments and support analysis and testing tasks, laying the groundwork for more senior profiles to complete the main improvements.
Focus the year mainly on learning and exploration, running tests and prototypes without a strong goal of direct product impact, leaving primary performance responsibility to others.

I may receive phone/text communications about career opportunities, interviews, job offers, and other related information from Talento Cientifico (+34 868 030 028 or +34 91 165 03 66) (Message and data rates may apply, message frequency varies). I may receive emails from talentocientifico.eu. I can opt-out at any time by unsubscribing or texting STOP. My data will be handled in accordance with the privacy policy and Terms of Service. To exercise your privacy rights, visit Privacy Rights & Opt-Out. For help, email support@loxo.co.