? Bengaluru · On-site · Full-time

Read this first

This is a programming job. You will write, debug, profile, and test production Python (and some C++) every working day. Computer vision is the domain; software engineering is the craft.

The first screening round is a 15-minute live coding session. It includes basic data structure concepts to advanced data structures like linked lists, stacks, queues, trees, graphs, etc. — this should be a warm-up for you, not an ambush.

What we do

We build high-precision measurement systems from multi-camera imagery: calibrated camera rigs, geometric reconstruction, and software that turns pixels into dimensions with accuracy claims we can defend.

What you'll own

•⁠ ⁠An existing measurement pipeline — multi-camera calibration, triangulation, and geometric computation.

•⁠ ⁠Performance: profiling NumPy-heavy numerical code and making it 10–100× faster without changing its answers.

•⁠ ⁠Validation: building test harnesses, synthetic-data checks, and regression suites.

•⁠ ⁠Deployment: packaging the pipeline to run reliably outside your laptop.

You must have

•⁠ ⁠Fluent Python — data structures, dictionaries, sets, complexity, and NumPy.





•⁠ ⁠Real debugging ability — experience hunting down difficult bugs.

•⁠ ⁠Comfort with git, code review, and writing maintainable code.

•⁠ ⁠Either hands-on geometric computer vision experience (calibration, stereo/multi-view, 3D reconstruction, OpenCV) or strong mathematical footing in linear algebra, optimisation, and statistics.

Nice to have

•⁠ ⁠C++

•⁠ ⁠Camera calibration or photogrammetry experience

•⁠ ⁠Industrial inspection / metrology exposure

•⁠ ⁠CUDA or serious performance work

•⁠ ⁠Scientific computing background

Explicitly not required

LLM, GenAI, RAG, or agent experience. This role contains none of it.

How we hire (and how quick)

•⁠ ⁠15-minute live coding screen — basics, camera on, any editor. Pass/fail, scheduled within 3 days of applying.

•⁠ ⁠Take-home — 4–6 hours inside a 48-hour window you choose. AI tools allowed; you'll defend your work live afterwards.

•⁠ ⁠Technical deep-dive — your take-home, past work, and live problem-solving.

- •⁠ ⁠Offer — every stage gets a yes/no within 3 working days. No ghosting, ever.

📌 Software Engineer (Bengaluru)
🏢 Swaya Studio
📍 Bengaluru