How to read this page
Who it's for. Third-year computer engineering students with general programming skills and no specialization. Every student receives a full Udacity subscription, so all programs listed are available at no extra cost. Pure "learn to program" courses are skipped; a language program appears only where it is a real prerequisite (scientific Python for the data paths, C++ for autonomous systems).
Hours. Udacity's listed hours are its own estimate for a focused, self-paced learner. Realistic hours are listed hours × 1.4, adding 40% for project work, debugging, review, and a concurrent college load. Example: a 40-hour program is 40 × 1.4 = 56 realistic hours, or about 2.3 weeks at 24 hours a week. Every path shows the addition written out so the numbers can be checked and updated.
Core and stretch. The core path is what a student must finish to graduate. It is front-loaded with the program that produces the most transferable portfolio project, so a student who stops at month three still leaves with something to show. The stretch block is for students who are ahead; it is never required.
Pace. At 24 h/week the core paths take 12–18 weeks. A student who can manage only 16–18 h/week during exam periods still finishes core in about six months. No path exceeds the nine-month ceiling even with the full stretch block.
| Path | Core listed | Core realistic | Core weeks | Core + stretch | Duration |
|---|---|---|---|---|---|
| Software Engineering + Agentic AI | 256 h | 358 h | 15 wk | 498 h | 21 wk / ≈ 4.8 mo |
| Cybersecurity + Cloud | 290 h | 406 h | 17 wk | 532 h | 23 wk / ≈ 5.1 mo |
| Data / Business Analytics | 208 h | 291 h | 13 wk | 410 h | 18 wk / ≈ 3.9 mo |
| Traditional Data Science / ML | 272 h | 381 h | 16 wk | 514 h | 22 wk / ≈ 4.9 mo |
| C++ / Autonomous Systems | 309 h | 433 h | 19 wk | 608 h | 26 wk / ≈ 5.9 mo |
Selecting students: the common gate
Interest is easy at twenty; you are selecting for follow-through. Every applicant, for every path, must complete the four steps below within a two-week window before admission. Each path then adds one task of its own, listed in its section.
- Finish a short free Udacity course (any relevant intro course) inside the two-week window. This alone removes roughly half of applicants.
- Submit a weekly schedule showing 20–24 hours blocked alongside the college timetable. The point is to see whether they have thought about it.
- Pass a 30-minute coding check in any language, confirming the programming baseline.
- Confirm reliable internet access in hours per week; the programs are video- and project-heavy.
Over-admit, then let the first deadline decide. Admit about 30% more students than the cohort can carry in month one, and let the first project milestone finalize the list. Announce this rule at admission so nobody feels cheated.
Follow-up: one structure, path-specific mentors
The mechanics of accountability are the same for every path; only the technical help differs. Run one shared system and attach one technical mentor per path.
- Weekly hours log. Self-reported, two minutes, every student.
- Bi-weekly 30-minute group check-in in peer groups of four to six per path. A peer to debug with is the cheapest retention tool available.
- Two missed consecutive check-ins trigger a one-to-one conversation, not a warning.
- Milestones are Udacity project submissions, with target weeks listed in each path. "Submitted project 2 by week 8" is measurable; "studied 24 hours" is not.
- One technical mentor per path reviews projects and answers questions. This is the only path-specific role.
- Traffic-light review, weekly. Green: on milestone. Yellow: one milestone behind. Red: two behind or absent. Look at yellow and red; ignore green.
- Public demo day at the end, announced on day one. Every student presents one project.
- Graduate = core path complete plus one portfolio project. Stretch content is a bonus, never a requirement.
Software Engineering + Agentic AI
Working with LLMs
Takes a student who can already program to a portfolio of shipped LLM-powered software. It front-loads a backend web project, the most transferable software-engineering artifact, then layers deep-learning literacy and generative-AI foundations before the capstone Agentic AI Nanodegree, whose human-reviewed multi-agent projects are the marquee portfolio pieces.
Core path required to graduate
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 1 | Backend Developer with Python | 49 h | Formerly the Full Stack Web Developer Nanodegree (nd0044). Ships a deployable Flask + SQL + Docker backend in the first three weeks, so every student leaves with a real software-engineering artifact before touching AI. |
| 2 | AI Programming with Python | 30 h (estimate) | Supplies the NumPy, pandas, PyTorch, and linear-algebra base that the generative-AI programs assume. Included because students know programming, not scientific Python. |
| 3 | Generative AI Fundamentals | 21 h | Conceptual on-ramp: foundation models, prompt engineering, and parameter-efficient fine-tuning. Short, and it makes the following Nanodegrees much faster to absorb. |
| 4 | Deep Learning | 45 h (estimate) | Neural-network depth (CNNs, RNNs, generative models in PyTorch) so that fine-tuning and RAG in the next program are understood rather than copied. |
| 5 | Generative AI | 56 h | Production generative-AI skills: fine-tuning with PEFT/QLoRA, a RAG chatbot, and a multimodal app using Hugging Face, Pydantic, and ChromaDB. Refreshed in 2026 from four courses to three. |
| 6 | Google Agentic AI Engineer | 55 h | The capstone. Builds a multi-agent travel assistant, a research agent, and a multi-agent system with Google's Agent Development Kit, Vertex AI, and the A2A protocol. Every project is human-reviewed. |
Stretch block optional, for students who are ahead
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 7 | Azure Generative AI Engineer | 40 h (estimate) | A second cloud provider's generative-AI stack (Azure OpenAI, RAG pipelines, GenAIOps) for students who want to be provider-agnostic. |
| 8 | AWS Machine Learning Engineer | 60 h (estimate) | Deployment and MLOps on SageMaker; turns a model builder into someone who can run models in production. |
Hours marked (estimate) could not be confirmed on the live Udacity page at the time of writing and should be re-checked before publishing.
Hours
Weekly plan
- Weeks 1–3Backend Developer with Python. Ship the backend portfolio project first.
- Weeks 4–5AI Programming with Python and Generative AI Fundamentals.
- Weeks 6–8Deep Learning.
- Weeks 9–11Generative AI. Fine-tuning and RAG projects.
- Weeks 12–15Google Agentic AI Engineer. Multi-agent capstone.
- Weeks 16–21Stretch: Azure Generative AI Engineer, then AWS Machine Learning Engineer.
Milestones
Each milestone is a Udacity project submission. Targets assume 24 h/week; at 16–18 h/week, shift each by roughly ×1.4.
- Week 3Backend app deployed
- Week 5Image classifier
- Week 8Deep-learning project
- Week 11RAG chatbot
- Week 15Multi-agent capstone
Before week 1
Python refresher before week 1. A two-hour HTTP/API refresher before the Agentic AI program. A basic linear-algebra brush-up helps with Deep Learning and Generative AI.
Considered and cut
- Full Stack JavaScript Developer — Duplicates web development already covered by the backend program; Python aligns better with the AI stack.
- AI for Business Leaders / Generative AI Fluency — Non-technical or redundant with Generative AI Fundamentals.
- AWS Machine Learning Engineer (in core) — Moved to stretch to keep the agentic-AI capstone as the finish line.
Who should join
All applicants pass the common gate described at the top of this page. Then, for this path:
Pre-admission task. Write a small script that calls an LLM API (any provider) and prints a structured response.
What it screens for. Self-directed documentation reading.
Follow-up
The shared follow-up structure applies, run by a software / agentic-AI mentor who reviews project submissions and answers technical questions.
Milestone weeks above are the yellow/red thresholds for the traffic-light review.
Graduate: core path complete plus one portfolio project presented on demo day. Stretch is never required.
Cybersecurity + Cloud
Defensive and offensive security on cloud infrastructure
A foundations-first security track that gives an early, tangible portfolio win (securing a business network), then builds cloud-architecture and DevOps depth, and finishes with hands-on offensive and defensive engineering. Front-loading the introductory Nanodegree means a student who stops at month three still holds completed security project artifacts.
Core path required to graduate
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 1 | Introduction to Cybersecurity | 69 h | The security foundation: threats, defenses, and incident response, with four projects including securing a business network. Placed first so the portfolio piece is banked early. |
| 2 | AWS Cloud Architect | 34 h | Teaches how cloud infrastructure is actually built (multi-account design, migration, cost) so that securing it later makes sense. |
| 3 | Cloud DevOps Engineer | 60 h (estimate) | Infrastructure as code, CI/CD, and Kubernetes. Modern security work lives in pipelines and clusters, and this program provides that fluency. |
| 4 | Ethical Hacker | 37 h | The offensive view: vulnerability management and a full penetration-test plan and report. Understanding attacks makes the final defensive program much more concrete. |
| 5 | Security Engineer | 90 h | The capstone: systems and infrastructure security, SIEM, and breach detection across four projects. The most job-relevant credential in the track. |
Stretch block optional, for students who are ahead
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 6 | Enterprise Security | 45 h (estimate) | Zero-trust and perimeter security on Azure for students who want a second cloud and an enterprise focus. |
| 7 | Security Architect | 45 h (estimate) | Risk frameworks and secure architecture design; the step from engineer to architect. |
Hours marked (estimate) could not be confirmed on the live Udacity page at the time of writing and should be re-checked before publishing.
Hours
Weekly plan
- Weeks 1–5Introduction to Cybersecurity. Bank the early security project.
- Weeks 6–8AWS Cloud Architect.
- Weeks 9–12Cloud DevOps Engineer.
- Weeks 13–15Ethical Hacker.
- Weeks 16–22Security Engineer.
- Weeks 23+Stretch: Enterprise Security, then Security Architect.
Milestones
Each milestone is a Udacity project submission. Targets assume 24 h/week; at 16–18 h/week, shift each by roughly ×1.4.
- Week 5Secure-a-network project
- Week 8AWS organization infrastructure
- Week 12CI/CD and IaC capstone
- Week 15Penetration-test report
- Week 22Security-engineering capstone
Before week 1
Linux command-line basics and networking fundamentals (OSI model, DNS, subnetting) before week 1. Udacity's free Linux Command Line Basics course is a good primer.
Considered and cut
- Security Analyst — Overlaps heavily with Security Engineer and Introduction to Cybersecurity.
- DevOps Engineer for Microsoft Azure — Kept a single-cloud spine on AWS.
- AWS Solutions Architect Associate prep — Exam preparation rather than build-focused.
Who should join
All applicants pass the common gate described at the top of this page. Then, for this path:
Pre-admission task. Complete a beginner free lab challenge (a basic capture-the-flag or intro security exercise) and write a one-page explanation of what you did.
What it screens for. Patience and methodical work.
Follow-up
The shared follow-up structure applies, run by a cybersecurity / cloud mentor who reviews project submissions and answers technical questions.
Milestone weeks above are the yellow/red thresholds for the traffic-light review.
Graduate: core path complete plus one portfolio project presented on demo day. Stretch is never required.
Data / Business Analytics
From messy data to decisions
A communication-first analytics track. It opens with Business Analytics with Tableau (no prerequisites, immediate dashboards to show), builds Python and SQL fluency, then advances through the Data Analyst Nanodegree and BI tooling. This is the lightest core path; the stretch block is where ambitious students should go rather than an optional extra.
Core path required to graduate
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 1 | Business Analytics with Tableau | 40 h | No prerequisites and three projects, so a dashboard portfolio piece exists by week four. Covers spreadsheets, SQL, and Tableau, the everyday tools of the job. |
| 2 | Programming for Data Science with Python | 45 h (estimate) | Students know programming but not the data stack. This builds pandas, NumPy, SQL, and Git habits used in every later program. |
| 3 | SQL | 40 h (estimate) | SQL is the single most-tested skill in analytics hiring. Joins, window functions, and data cleaning at a depth the introductory programs don't reach. |
| 4 | Data Analyst | 43 h | The core credential: data wrangling, exploratory analysis, and visualization across three projects using the Python stack built above. |
| 5 | Data Analysis and Visualization with Microsoft Power BI | 40 h (estimate) | Power BI is the most common BI tool in regional employers. Modeling and DAX round out the Tableau skills from the first program. |
Stretch block optional, for students who are ahead
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 6 | Business Intelligence Analytics | 40 h (estimate) | The full BI life cycle and reporting practice for students heading toward BI roles. |
| 7 | Data Scientist | 45 h (estimate) | An on-ramp into machine learning for analysts who want to move toward data science. |
Hours marked (estimate) could not be confirmed on the live Udacity page at the time of writing and should be re-checked before publishing.
Hours
Weekly plan
- Weeks 1–4Business Analytics with Tableau. Deliver a dashboard early.
- Weeks 5–7Programming for Data Science with Python.
- Weeks 8–9SQL.
- Weeks 10–12Data Analyst.
- Weeks 13–14Power BI.
- Weeks 15–17Stretch: Business Intelligence Analytics, then Data Scientist.
Milestones
Each milestone is a Udacity project submission. Targets assume 24 h/week; at 16–18 h/week, shift each by roughly ×1.4.
- Week 4Tableau dashboard
- Week 7Python data project
- Week 9SQL analysis
- Week 12Data Analyst capstone
- Week 14Power BI report
Before week 1
None strictly required. A short Python refresher accelerates the Python program; basic statistics helps with Data Analyst.
Considered and cut
- Predictive Analytics for Business — Built on Alteryx, which is niche compared with the Python/SQL/Power BI spine.
- Data Engineering with AWS — Belongs to a heavier engineering path.
- Marketing Analytics — Domain-specific rather than general analytics.
Who should join
All applicants pass the common gate described at the top of this page. Then, for this path:
Pre-admission task. From a messy CSV, produce three findings and one chart.
What it screens for. Clarity of communication, not code quality.
Follow-up
The shared follow-up structure applies, run by a data / business-analytics mentor who reviews project submissions and answers technical questions.
Milestone weeks above are the yellow/red thresholds for the traffic-light review.
Graduate: core path complete plus one portfolio project presented on demo day. Stretch is never required.
Traditional Data Science / ML
Statistics, models, and production
A rigor-first machine-learning track. It secures Python and SQL with an early Data Analyst portfolio project, then moves through the mathematics of AI programming, supervised and unsupervised learning in PyTorch, the full Data Scientist Nanodegree, and a production ML-engineering capstone. This is the most prerequisite-sensitive track, so selection is strict.
Core path required to graduate
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 1 | Data Analyst | 43 h | Front-loaded portfolio project. Data wrangling and exploratory analysis are the daily work of data science, and this gets them done first. |
| 2 | Programming for Data Science with Python | 45 h (estimate) | Builds the pandas, SQL, and Git fluency every later program assumes. |
| 3 | AI Programming with Python | 30 h (estimate) | The mathematics: linear algebra, calculus, and NumPy/PyTorch, delivered as code. The bridge between analysis and modeling. |
| 4 | Introduction to Machine Learning with PyTorch | 49 h | Supervised learning, neural networks, and unsupervised learning with three projects. Chosen over the TensorFlow variant to keep one framework across the track. |
| 5 | Data Scientist | 45 h (estimate) | Pipelines, experiment design, recommendation systems, and an open capstone. The credential that names the role. |
| 6 | AWS Machine Learning Engineer | 60 h (estimate) | Deploying and operating models on SageMaker. Separates candidates who can train a model from those who can ship one. |
Stretch block optional, for students who are ahead
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 7 | Deep Learning | 45 h (estimate) | CNNs, RNNs, and generative models for students heading into vision or sequence work. |
| 8 | Natural Language Processing | 50 h (estimate) | Text processing, embeddings, and sequence models; broader utility than reinforcement learning, which was the alternative. |
Hours marked (estimate) could not be confirmed on the live Udacity page at the time of writing and should be re-checked before publishing.
Hours
Weekly plan
- Weeks 1–3Data Analyst. Front-loaded portfolio project.
- Weeks 4–6Programming for Data Science with Python.
- Weeks 7–8AI Programming with Python. Math and PyTorch foundations.
- Weeks 9–11Introduction to Machine Learning with PyTorch.
- Weeks 12–14Data Scientist.
- Weeks 15–18AWS Machine Learning Engineer.
- Weeks 19–21Stretch: Deep Learning, then Natural Language Processing.
Milestones
Each milestone is a Udacity project submission. Targets assume 24 h/week; at 16–18 h/week, shift each by roughly ×1.4.
- Week 3Data Analyst project
- Week 6Python data project
- Week 8Image classifier
- Week 11Supervised-learning project
- Week 14Data Scientist capstone
- Week 18ML deployment capstone
Before week 1
Linear algebra, probability, and derivatives refresher before week 7; the PyTorch programs assume comfort with these. Keep a Python data-manipulation warm-up in weeks 1–2.
Considered and cut
- Introduction to Machine Learning with TensorFlow — Standardized on PyTorch to avoid a duplicate framework.
- Machine Learning DevOps Engineer — Folded into the AWS Machine Learning Engineer choice.
- Deep Reinforcement Learning — Considered for stretch; NLP has broader utility. An easy swap if a cohort prefers it.
Who should join
All applicants pass the common gate described at the top of this page. Then, for this path:
Pre-admission task. A short math check (linear algebra, probability, derivatives) plus a small Python data task.
What it screens for. Mathematical readiness. Be strict; prerequisites decide outcomes here.
Follow-up
The shared follow-up structure applies, run by a data-science / ML mentor who reviews project submissions and answers technical questions.
Milestone weeks above are the yellow/red thresholds for the traffic-light review.
Graduate: core path complete plus one portfolio project presented on demo day. Stretch is never required.
C++ / Autonomous Systems
Perception, robotics, and self-driving software
A hardware-and-robotics track anchored in C++. It starts with the C++ Nanodegree (route-planner and system-monitor projects to show early), adds the self-driving-car on-ramp, then builds perception, robotics with ROS, and the flagship Self-Driving Car Engineer capstone. This is the heaviest track; persistence matters more than breadth, so admit fewer students.
Core path required to graduate
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 1 | C++ | 42 h | Included as a language program because everything downstream is C++. Four projects (OpenStreetMap route planner, system monitor, memory-managed chatbot, concurrency) give early portfolio pieces. |
| 2 | Introduction to Self-Driving Cars | 45 h (estimate) | Bridges Python to C++ and introduces the linear algebra and object-oriented patterns of autonomous vehicles. De-risks the jump into Sensor Fusion. |
| 3 | Sensor Fusion | 63 h | Perception: lidar obstacle detection, camera, radar, and Kalman filters across five projects. The core technical skill of the field. |
| 4 | Robotics Software Engineer | 81 h | ROS, Gazebo, localization, SLAM, and path planning with five projects. Turns perception into a working robot. |
| 5 | Self-Driving Car Engineer | 78 h | The capstone: computer vision, sensor fusion, localization, and control on the Waymo dataset and CARLA simulator, seven projects. |
Stretch block optional, for students who are ahead
| Program | Udacity hours | Why it's here | |
|---|---|---|---|
| 6 | Deep Learning | 45 h (estimate) | Deepens the neural-network side of perception for students who want a vision specialization. |
| 7 | Flying Car and Autonomous Flight Engineer | 80 h (estimate) | Aerial planning, controls, and estimation; the same discipline applied to drones. |
Hours marked (estimate) could not be confirmed on the live Udacity page at the time of writing and should be re-checked before publishing.
Hours
Weekly plan
- Weeks 1–3C++. Bank the route-planner and system-monitor projects.
- Weeks 4–6Introduction to Self-Driving Cars.
- Weeks 7–11Sensor Fusion.
- Weeks 12–17Robotics Software Engineer.
- Weeks 18–23Self-Driving Car Engineer. Capstone.
- Weeks 24+Stretch: Deep Learning, then Flying Car.
Milestones
Each milestone is a Udacity project submission. Targets assume 24 h/week; at 16–18 h/week, shift each by roughly ×1.4.
- Week 3C++ route planner
- Week 6AV math and OOP project
- Week 11Sensor-fusion tracking
- Week 17Home-service robot (ROS)
- Week 23Self-driving-car capstone
Before week 1
C++ pointers and classes refresher plus Linux basics before week 1. Linear algebra and calculus (derivatives, matrices) are essential before Sensor Fusion and Robotics. Reserve buffer weeks during exam periods.
Considered and cut
- Deep Reinforcement Learning — Overlaps with perception content and adds load.
- Flying Car (in core) — Kept in stretch so the ground-vehicle capstone is the finish line.
- Skipping Introduction to Self-Driving Cars — Rejected; it bridges Python to C++ math and lowers the failure rate in Sensor Fusion.
Who should join
All applicants pass the common gate described at the top of this page. Then, for this path:
Pre-admission task. A C++ task using pointers and classes, plus a short statement of interest in hardware or robotics.
What it screens for. Persistence. Select fewer, more stubborn students.
Follow-up
The shared follow-up structure applies, run by a C++ / robotics mentor who reviews project submissions and answers technical questions.
Milestone weeks above are the yellow/red thresholds for the traffic-light review.
Graduate: core path complete plus one portfolio project presented on demo day. Stretch is never required.