Research Engineer / Applied Scientist

Mathpix; United States; On-site; USD 90,000 - 120,000; Full time

Job summary

Mathpix is seeking a Research Engineer / Applied Scientist to design and run large-scale computational experiments. Ideal candidates hold a PhD in physics or related fields, and possess strong programming skills in Python and C++, with experience managing complex research projects.

The position allows for growth within document understanding problems, collaborating with ML and engineering teams. Applicants with experience in deep learning, HPC clusters, and publications in their field are encouraged to apply.

Qualifications

• PhD or equivalent research experience in a quantitative discipline.
• Strong programming skills in Python and/or C++.
• Experience managing computational research projects end-to-end.

Responsibilities

• Design and run large-scale computational experiments to improve ML systems.
• Implement performance-critical components in Python and C++.
• Investigate model failure modes and pipeline edge cases.

Skills

Strong programming skills on Linux in Python and/or C++

Multi-month computational research project management

Ability to self-teach difficult technical material

Education

PhD in physics or another quantitative discipline

Tools

Linux

PyTorch

Job description

Location: Brooklyn, NY or Bay Area preferred

Mathpix is looking for a Research Engineer / Applied Scientist to join our team. We believe the hardest part of being a great applied scientist isn't knowing the latest model architectures — it's being able to design and run long-horizon computational experiments, debug from first principles, and teach yourself whatever you need to know to make progress on a hard problem. Those are exactly the skills a strong PhD already has. We'd rather hire that mindset and bring you up to speed on modern ML and computer vision than hire the marginal ML PhD.

We’re especially interested in PhDs from physics, applied math, computational chemistry, astronomy, computational biology, neuroscience, electrical engineering, mechanical engineering, and adjacent quantitative disciplines. If you spent years running simulations on a cluster, writing performance‑critical Python or C++ on Linux, and self‑teaching whatever technique your thesis required — we want to talk to you, whether or not you’ve done ML before. You’ll work on real problems in document understanding alongside our ML and engineering teams, with room to grow into whichever parts of the stack you find most interesting.

Responsibilities

• Design and run large‑scale computational experiments to evaluate and improve our ML systems and document processing pipelines
• Implement performance‑critical components in Python and/or C++ on Linux
• Investigate model failure modes and pipeline edge cases from first principles, using whatever tools the problem requires
• Pick up unfamiliar techniques — new model architectures, optimization methods, evaluation methodologies — and apply them to production problems
• Collaborate with the ML team on data curation, training, and evaluation, and with engineering on integrating results into production

Required skills

• PhD in physics or another quantitative discipline (applied math, computational chemistry, astronomy, computational biology, EE, ME, etc.), or equivalent research experience
• Strong programming skills on Linux in Python and/or C++
• Track record managing a multi‑month computational research project end‑to‑end — designing experiments, running them, debugging, iterating, drawing conclusions
• Demonstrated ability to self‑teach difficult technical material

Nice to have

• Exposure to deep learning, especially in PyTorch
• Experience with HPC clusters, GPUs, or distributed compute
• Publications in your field
• Background in OCR, computer vision, NLP, or document understanding
• Open‑source contributions, technical writing, or other work shared publicly

If you believe you would be a good fit for this role, please apply using the link below.