The Course Handbook is a PDF you can keep: course facts, learning outcomes, the full week by week syllabus, assessment weightings and the spoken defence. Or print this page straight from your browser for the same layout.
Why this course exists
Build basic deep learning models in TensorFlow. At Forward this is taught from inside an institution that runs on these systems every day, so the course teaches the working practice rather than the marketing. It sits in Machine Learning and Deep Learning, aimed at analysts and developers going deeper. You are expected to be comfortable with the tools of your own trade. We move at working pace.
Who it is for
· Analysts and developers going deeper
· People already doing this work who want it done properly
· Anyone who needs evidence of capability, not a certificate of attendance
· Forward University and Forward College applicants building a case for recognition of prior learning
Not for you if
· You want a certificate without building anything.
· You cannot commit 2 hours a week for 2 weeks.
· You are looking for an NCHE-accredited award. This is a professional micro-credential, and we say so in public.
· You already do this professionally and want a research-level treatment.
Entry: No formal entry requirements, but you will get far more out of this if you have already used AI tools for real work, or completed an Introductory Academy course.
What you will be able to do
01Explain, in your own words and to a sceptical colleague, what this course covers and where it stops.
02Work confidently with framing the problem.
03Work confidently with writing it up.
04Produce the course project: A trained or evaluated model with an honest write-up of what it gets wrong.
05Judge when to use these methods, when to refuse, and who remains accountable for the output.
06Carry the work forward into machine learning and deep learning at Forward, or into a longer path.
The full syllabus, week by week
Every week has one live session, self-paced material, and a build task that becomes part of your final project. Nothing here is a reading list.
Week 1
Framing the problem
Prediction versus explanation, the target variable and what success would look like.
You do
Write the problem statement, the metric and the baseline to beat.
Checkpoint
A short build task, marked against the published rubric before the next live session.
Week 2
Writing it up
Reporting a model so someone else could trust, reproduce or refuse it.
You do
Finish the model card and prepare the defence.
Checkpoint
Project submitted, then the spoken defence with a human assessor.
Save the whole syllabus and the learning outcomes as a PDF, or print this page.
The project, and how it is marked
You build: A trained or evaluated model with an honest write-up of what it gets wrong.
30%
Weekly build tasks
Small pieces of the project, submitted each week and marked against a published rubric. Late is fine; missing is not.
50%
The course project
A trained or evaluated model with an honest write-up of what it gets wrong.
20%
Spoken defence
About fifteen minutes with a human assessor, who asks you why you made the choices you made. Mastery, not averaging: you may re-sit the defence.
The spoken defence
· Fifteen minutes, online, with a human assessor.
· You show the thing you built and answer why, not what.
· Mastery standard: if you do not meet it, you re-sit rather than average out.
· The result and the assessor are recorded on your verified credential.
How Musomesa helps
· Available every hour of the week, and it has read the same course material you have.
· It will not hand you the answer to a graded task. It will ask you what you have tried.
· It flags when your reasoning is confident and wrong, which is the whole point of the course.
· Your instructor sees the same progress signals, so the live session starts where you actually are.
What you need
PyTorch or TensorFlow
GPU notebook environment
A general-purpose AI assistant
Musomesa, your Forward AI tutor
Who teaches FA-1057, and how to get help
FA-1057 is taught by a small team rather than one overloaded lecturer. This is an intermediate course, so your lead instructor teaches from current practice rather than slides. Your named team is confirmed by email before the cohort opens; the roles and the hours below are fixed.
Lead Instructor
A working practitioner in machine learning and deep learning, teaching this course alongside their own practice.
· Runs the weekly live session and answers questions in it
· Sets and reviews the weekly build task
· Holds one evening office hour every week of the course
Promise: Reads and responds to every question raised in the live session or the cohort channel.
Studio Coach
A Forward coach attached to your cohort for the whole course, not per session.
· One-to-one help when a build task stalls
· Weekend office hour for learners in other timezones or on shift work
· Keeping your project on track for the 2-week deadline
Promise: If you are stuck for more than 48 hours, the coach reaches out to you first.
Assessment Examiner
An examiner who did not teach you, so marking is independent of teaching.
· Marks the project against the published weightings
· Runs your short spoken defence
· Issues the verified micro-credential to your Achievement Wallet
Promise: Written feedback with every mark, and a stated route to a second opinion.
Musomesa, your AI tutor
Available on this course at any hour, on top of the human team above, never instead of it.
· Answers course questions between sessions, in context of this syllabus
· Walks you through worked examples at your pace
· Flags to your Studio Coach when you are repeatedly stuck
Promise: Musomesa never marks your work and never issues your credential. People do that.
Office hours
When
Time
Who
Format
Every Tuesday
20:00 – 21:30 EAT
Lead Instructor
Open video room. Bring your build task, no agenda needed, drop in and out.
Every Saturday
08:30 – 10:00 EAT
Studio Coach
Bookable 20-minute one-to-one slots for anything you would rather not ask in a group.
Any day
24 hours
Musomesa
Chat tutor inside the course, grounded in this syllabus and your own submitted work.
Defence week
Booked with you
Assessment Examiner
A short spoken defence of your project, scheduled around your timezone in week 2.
All times are East Africa Time (UTC+3). If you are outside East Africa, tell your Studio Coach at enrolment and the weekend slot is moved into your working day, not ours.
How to get help
Cohort channel
Course questions, tool problems, sharing progress with the rest of the cohort.
Response: Same day on weekdays, from an instructor or coach.
Musomesa, in-course tutor
Explanations, worked examples, unblocking a build task at 2am.
Response: Immediate, any hour.
Office hours
Anything that needs a person looking at your actual work with you.
Response: Two working days, with the reference from your enrolment.
If you get stuck, in order
01Ask Musomesa first. Most blockers are explained in under a minute.
02Still stuck after a day? Post in the cohort channel so the instructor and your peers both see it.
03Still stuck at the next office hour? Bring the work itself, not a summary of it.
04Unhappy with a mark or the course? Email Academy support and ask for the Assessment Examiner's second-opinion route. It is a published process, not a favour.
Named instructors, their biographies and cohort-specific hours are published with each cohort as it opens. We do not list staff on a course page before they are contracted to teach it.
Questions people ask about FA-1057
Are there any prerequisites for FA-1057?
No formal qualifications are required for Academy courses. No formal entry requirements, but you will get far more out of this if you have already used AI tools for real work, or completed an Introductory Academy course.
How much time does the course take each week?
Intro to Deep Learning with TensorFlow runs for 2 weeks at 2 hours a week. Each week has one live session, self-paced material, and a build task that becomes part of your final project.
How is the course assessed?
There are no exams. You are assessed on weekly build tasks (30%), the course project (50%), spoken defence (20%), including a short spoken defence of the project you build.
How does Musomesa support me on the course?
Musomesa, Forward's AI tutor, works alongside your human instructor. Available every hour of the week, and it has read the same course material you have.
What does FA-1057 cost?
This course is free. No payment is taken at any point.
Learning paths that include FA-1057
A path is three courses taken in order with one combined Statement of Capability at the end. These are the paths recommended for a intermediate course in this track.
18 weeks · next step
The Analyst Path
For: Officers, coordinators and managers who live in spreadsheets
Stacks this intermediate work into a full path with a combined Statement of Capability.
