Associate ML Engineer in New Delhi, Delhi at Wadhwani AI

New
Wadhwani AI
New Delhi, Delhi, India
Posted on
New job! Apply early to increase your chances of getting hired.

Job Description

About the role

We build AI products for public health and early grade education, deployed in real field conditions across sub-Saharan Africa and South Asia. The products are used by data collectors, health workers, and teachers, often on low cost devices, often without reliable connectivity, and always under data protection obligations that vary by country. Our models work on speech, audio, and vision, on data that is noisy and underrepresented in public datasets.

You will work on these models from data preparation through training and evaluation to handing a model over for deployment. The team is small, so you will work on the whole pipeline rather than one stage of it. We are looking for someone with solid fundamentals who wants to learn what it takes to make a model work outside a notebook.


What you will do

* Prepare and clean field data for training: audio, images, and their annotations, including quality checks and dataset splits.

* Run training and fine tuning experiments on speech, audio, and vision models, track them properly, and report what worked and what did not.

* Build evaluation sets and metrics that reflect how the model will actually be used, not just a held out test split.

* Run error analysis: listen to the audio, look at the images, and find the patterns behind the failures.

* Write clean, tested Python for data pipelines and inference code that an engineer can deploy.

* Build internal annotation and review tools with the engineering team when the project needs them.

* Document experiments, datasets, and decisions so that others can reproduce your work.

* Follow our data protection practices when handling health and child data, and flag anything that looks off.


What we are looking for

Required

* 0 to 3 years of experience, or a strong record of ML projects you built yourself. Internships, research projects, Kaggle work, and open source contributions count.

* Solid Python, with PyTorch or a comparable deep learning framework, and comfort with NumPy and pandas.

* Sound understanding of ML fundamentals: training and validation, overfitting, loss functions, and how to read a learning curve.

* At least one project involving speech, audio, or computer vision that you can walk us through end to end.

* Basic familiarity with Git, the command line, and running jobs on a remote GPU machine.

* Clear written communication, since most coordination is asynchronous and across time zones.

* Curiosity and rigour: you check your results before believing them and ask for help early when stuck.

Nice to have

* Experience fine tuning pretrained speech models such as Whisper or wav2vec, or vision and OCR models.

* Exposure to model serving, quantisation, or running inference on CPU or mobile.

* Experience with experiment tracking tools and reproducible training setups.

* Familiarity with low resource language work, child speech, or data collected in field conditions.

* Interest in health or education technology, or in working on products deployed in low resource settings.

Explore Related Opportunities

Job Location

New Delhi, Delhi, India

Frequently asked questions about this position

Apply For This Position

Apply Now