Machine Learning Tutor Job — Remote, Freelance, Rs 500-1,500/hr

The Machine Learning tutor job at MEB involves running 1:1 live online sessions and providing homework guidance within those sessions, mainly for students in the USA and the Gulf. Most applicants are graduate or advanced undergraduate students enrolled in data science, computer science, or quantitative finance programmes who arrive with specific model-building or implementation questions rather than broad conceptual gaps. Sessions frequently centre on debugging code, interpreting model outputs, and working through mathematical derivations — all in real time, on a shared digital whiteboard with a pen tablet. The role demands fluency across both the theoretical and the applied sides of the subject: a tutor who can derive the backpropagation update rule and also fix a TensorFlow shape mismatch in the same hour.

What the role involves

• Running 1:1 live online sessions on Machine Learning topics ranging from foundational probability and linear algebra prerequisites through to deep learning architectures and model deployment.
• Guiding students through their own problem sets — explaining the method and the reasoning, not supplying finished answers.
• Working through mathematical derivations on a shared digital whiteboard, using a pen tablet, at the pace a student can follow.
• Helping students understand why a model is underperforming: diagnosing bias-variance problems, regularisation choices, data-leakage issues, and similar failure modes.
• Keeping sessions focused and on time; most are 60 minutes or less and students arrive with a specific problem already in mind.

Topics you will be expected to teach

• Supervised learning: linear regression, logistic regression, decision trees, random forests, gradient boosting (XGBoost, LightGBM)
• Unsupervised learning: k-means clustering, hierarchical clustering, DBSCAN, dimensionality reduction
• Principal Component Analysis (PCA) and Singular Value Decomposition (SVD)
• Bias-variance tradeoff, cross-validation, and model selection
• Regularisation: L1 (Lasso), L2 (Ridge), Elastic Net, dropout
• Neural networks: feedforward architectures, backpropagation, activation functions, optimisers (SGD, Adam, RMSProp)
• Convolutional Neural Networks (CNNs) for image tasks
• Recurrent Neural Networks (RNNs), LSTMs, and sequence modelling
• Transformer architectures and attention mechanisms
• Support Vector Machines (SVMs) and kernel methods
• Bayesian methods: naive Bayes, Gaussian processes, Bayesian optimisation
• Evaluation metrics: precision, recall, F1, ROC-AUC, confusion matrices, calibration
• Feature engineering, feature selection, and handling imbalanced data
• Model deployment fundamentals: pipelines, serialisation, and basic MLOps concepts

A problem you should be able to solve

You have a binary classification dataset with 10,000 samples and 50 features. After training a logistic regression model with no regularisation, the training accuracy is 96% and the validation accuracy is 71%. Describe in precise terms what is happening, derive the form of the cost function you would use to address it, state how you would choose the regularisation strength, and explain what the optimal solution looks like geometrically in weight space.

If you cannot set this up and solve it in under five minutes without looking anything up, this role is not the right fit.

Who we are looking for

Subject mastery

Machine Learning sits at the intersection of statistics, linear algebra, optimisation, and software engineering. You need to be genuinely comfortable in all four areas. That means deriving gradient descent updates from first principles, explaining the kernel trick without hand-waving, and reading a model’s learning curves to diagnose training problems — not just knowing which scikit-learn class to import. Students in graduate programmes will probe your understanding quickly; surface-level familiarity does not survive a 60-minute session with a motivated MSc student.

Speed and accuracy under deadline

Sessions are live and unscripted. A student may arrive with a derivation they cannot close, a failing training loop, or a result from an experiment that contradicts what their course notes predict. You need to identify the issue, reason through it correctly, and explain it clearly — all within the session window. If you need to look up basic formulas or run test code before answering standard questions, sessions will stall and the student will not return.

Education and background

A degree in Computer Science, Statistics, Mathematics, Electrical Engineering, or a closely related field from IIT, IISc, ISI, NIT, or an institution of equivalent rigour is the baseline expectation. Candidates with a strong research background — published work, a competitive ML track record, or substantial industry experience building production models — will also be considered. The subject is taught at graduate level; your background needs to match or exceed that level.

Setup, availability and communication

You will need a reliable laptop, stable broadband, a working camera and microphone, and a pen tablet — without exception. Sessions are conducted online on a shared digital whiteboard; handwriting legibility and drawing speed matter. Most student requests arrive between 5 PM and 9 AM IST, which reflects USA and Gulf time zones. Your English must be clear and precise; students are almost entirely non-Indian, and any ambiguity in explanation costs them time they do not have.

Do not apply if

• You need a guaranteed monthly income or a fixed number of hours each week.
• You cannot work between 5 PM and 9 AM IST on at least one or two nights a week.
• You do not own a pen tablet and are not willing to acquire one before your first session.
• Your Machine Learning knowledge is limited to running pre-written notebooks or following tutorials without understanding the underlying mathematics.
• You expect to look up derivations, formulas, or standard algorithm behaviour during a live session.

What this job is not

This is not a salaried position. There is no monthly pay guarantee, no retainer, and no minimum number of hours. Work is offered as it arises, job by job, and you are free to accept or decline each assignment. This is also not a route to completing graded work on a student’s behalf; the role is to guide students to understand and solve problems themselves, and any tutor who crosses that line ends their engagement with MEB immediately. This is not a fixed-shift job and is not suitable as a primary income source for someone without financial flexibility.

Pay and payment terms

The pay rate for this role is Rs 500 – Rs 1,500 per hour. The exact rate for a given assignment depends on the level of the material, the complexity of the session, the deadline, and the type of work involved. The fee is agreed before any work begins; you will not be asked to start without knowing what you will be paid. You may accept or decline any assignment offered to you. Payment is made on time. There are no deductions, no platform commissions visible to you, and no delayed disbursements.

This is a freelance engagement. There is no guaranteed income, no fixed monthly minimum, and no retainer. Freshers are eligible only where subject depth is clearly exceptional.

How work is assigned at MEB

Student requests come in as they arise — there is no fixed schedule and no predictable volume. When a request matches your subject area and availability, MEB offers it to you. Work is distributed fairly among tutors who cover the same subject. You can accept or decline. If you accept, the session details and the agreed fee are confirmed before you begin. Most assignments are driven by USA and Gulf student demand, which means the bulk of offers arrive during the evening and night in Indian Standard Time.

Academic integrity rules for tutors

Tutors at MEB guide students to understand and solve problems independently. A tutor must not complete graded assessments, take-home exams, or any other work that is to be submitted for a mark on a student’s behalf. Explaining a method is permitted; producing a finished answer for submission is not.

Tutors must not share personal contact details — phone numbers, personal email addresses, or social handles — with students, and must not negotiate fees or book sessions outside the MEB platform. Any tutor found doing so ends their engagement with MEB immediately.

Full details are available in the MEB academic integrity policy.

Selection process

1. Submit your application through the tutoring jobs hub.
2. Shortlisting based on your subject background, degree, and relevant experience.
3. A subject-specific written test followed by a short mock session on a shared digital whiteboard — you will need your pen tablet for this step.
4. Onboarding, after which work is offered job by job as student requests come in.

For questions before applying, reach us on WhatsApp at +91 8971 383660 or by email at meb@myengineeringbuddy.com.

Questions from applicants

Do I need prior tutoring experience, or will strong academic credentials be enough?

Strong academic credentials in Machine Learning — particularly from IIT, IISc, ISI, NIT, or equivalent institutions — are the primary filter. Prior tutoring experience is valued but not a strict requirement if your subject depth is clearly at the level needed. Freshers with exceptional mastery are eligible. What matters most is whether you can explain complex ML concepts clearly and correctly to a graduate student under session pressure.

How many sessions can I expect per week, and is there a minimum commitment?

There is no guaranteed number of sessions and no minimum commitment on either side. Work depends entirely on student demand in your subject area and your availability during the main working window of 5 PM to 9 AM IST. In practice, active tutors who cover in-demand Machine Learning topics tend to receive one to three offers per week, but this is not guaranteed and varies by season and course cycles.

Will I be tested on mathematics and theory, or only on practical coding ability?

The selection test covers both. Machine Learning at the level MEB’s students require demands fluency in the underlying mathematics — probability, linear algebra, optimisation, and statistical learning theory — as well as the ability to read, write, and debug code in Python using standard ML libraries. A candidate who is strong in one area but weak in the other is unlikely to pass the test, because students regularly arrive with questions that span both.

Is a pen tablet strictly required, or can I use a mouse or trackpad for the whiteboard?

A pen tablet is strictly required. Sessions involve working through derivations, sketching model diagrams, and annotating equations in real time on a shared digital whiteboard. Writing with a mouse or trackpad produces output that is slow and illegible under session conditions. Candidates who do not own a pen tablet will need to acquire one before the mock session step in the selection process.

How does MEB handle situations where a student’s question falls into a grey area of academic integrity?

The guiding principle is straightforward: explain the method, do not produce a finished answer for submission. If a student asks you to walk through a concept, help them understand an algorithm, or debug their own code, that is within scope. If a student asks you to complete a graded problem set, a take-home exam, or any assessment that will be submitted under their name, decline and report it through the MEB platform. When in doubt, decline the specific request and explain why — the academic integrity policy provides clear guidance, and MEB’s team is available if a situation is genuinely ambiguous.

Related tutoring job openings

• Power BI tutor job
• Econometrics tutor job
• Statistics tutor job
• Probability tutor job
• Data Science tutor job

Looking for tutoring rather than a job? Visit our Machine Learning tutor page.