CS511 Advanced Artificial Intelligence
Syllabus | International University of Sarajevo - Last Update on Oct 10, 2026
Syllabus Quick Jump
Search and navigate to any syllabus instantly
Computer Sciences and Engineering
Ali Almisreb
Course Lecturer
Course Objectives
This graduate course starts where AID401 Deep Learning ends (the Transformer, CNNs, RL basics) and studies the three pillars of current AI research in technical depth: large language models (LLMs), vision-language models (VLMs) and LLM-based agents. Every topic is developed with equations, derivations and worked numeric examples, implemented in weekly PyTorch / Hugging Face labs on open-weight models, and connected to recent research papers. Students present a research paper and carry out a semester-long, reproducible research project. Learning outcomes. On successful completion, students will be able to:
- Derive and analyse modern LLM architectures and pretraining: tokenization, RoPE, GQA, scaling laws, distributed training (ZeRO/FSDP), FlashAttention, long context and mixture-of-experts.
- Implement and evaluate post-training and inference methods: SFT and LoRA/QLoRA, reward models, RLHF/PPO and DPO, reasoning with GRPO and test-time compute, quantization and speculative decoding.
- Explain, build and evaluate vision-language models: contrastive pretraining (CLIP/SigLIP), connectors, grounding, hallucination, and unified multimodal generation with discrete visual tokens.
- Design, build and evaluate LLM agents with retrieval (RAG), memory and tools (function calling, MCP), and analyse their reliability and security (prompt injection, adversarial attacks).
- Critically read current research and conduct a reproducible research project with sound experimental design and statistical reporting, presented as a paper and a talk.
Learning Outcomes
After successful completion of the course, the student will be able to:
Course Materials
Required Textbook
Additional Literature
Teaching Methods
Each week combines a technical lecture with a hands-on lab: concepts are first derived mathematically and checked on small numeric examples, then implemented from scratch and with standard libraries, and finally evaluated on real models and benchmarks
Students read the assigned papers before class, each student presents and leads the discussion of one recent research paper, and teams of 1–3 students carry out a semester-long research project on LLMs, VLMs or agents, reported as a workshop-style paper with a reproducibility package
Weekly Topics
| Week | Topic | Readings / References |
|---|---|---|
| 1 | Modern LLM Architecture and Scaling | Vaswani et al. (2017); Hoffmann et al. (2022); Raschka (2024) Ch. 2–4 |
| 2 📍 | Pretraining at Scale: Data and Distributed Training | Rajbhandari et al. (2020) ZeRO; Shoeybi et al. (2019) Megatron-LM |
| 3 | Long Context, Fast Attention and Mixture of Experts | Dao et al. (2022) FlashAttention; Peng et al. (2024) YaRN; Fedus et al. (2022) Switch Transformer |
| 4 | Post-training I: SFT and Parameter-Efficient Fine-Tuning | Hu et al. (2022) LoRA; Dettmers et al. (2023) QLoRA; Zhou et al. (2023) LIMA |
| 5 | Post-training II: RLHF and DPO | Ouyang et al. (2022) InstructGPT; Schulman et al. (2017) PPO; Rafailov et al. (2023) DPO |
| 6 | Reasoning and Test-Time Compute | Wei et al. (2022); Wang et al. (2023); Shao et al. (2024) GRPO; DeepSeek-AI (2025) DeepSeek-R1 |
| 7 | LLM Inference and Serving | Kwon et al. (2023) vLLM; Frantar et al. (2023) GPTQ; Leviathan et al. (2023) |
| 8 | Midterm Week: LLM Review and Research Proposal Presentations | Lecture notes and labs of Weeks 1–7 |
| 9 | Vision-Language Model Architectures | Radford et al. (2021) CLIP; Zhai et al. (2023) SigLIP; Liu et al. (2023) LLaVA; Wang et al. (2024) Qwen2-VL |
| 10 | Grounding, Perception and VLM Evaluation | Minderer et al. (2023) OWLv2; Li et al. (2023) POPE; Leng et al. (2024) VCD |
| 11 | Unified Multimodal Models | van den Oord et al. (2017) VQ-VAE; Chang et al. (2022) MaskGIT; Mentzer et al. (2024) FSQ; Zhou et al. (2024) Transfusion |
| 12 | Retrieval and Memory | Karpukhin et al. (2020) DPR; Lewis et al. (2020) RAG; Khattab & Zaharia (2020) ColBERT |
| 13 | LLM and Multimodal Agents | Yao et al. (2023) ReAct; Schick et al. (2023) Toolformer; Anthropic (2024) MCP specification |
| 14 | Training, Evaluating and Securing Agents | Yao et al. (2024) tau-bench; Greshake et al. (2023); Zou et al. (2023) GCG; Debenedetti et al. (2025) CaMeL |
| 15 | Course Synthesis and Final Project Presentations | Lecture notes and labs of Weeks 9–14 |
Course Schedule (All Sections)
| Section | Type | Day 1 | Venue 1 | Day 2 | Venue 2 |
|---|---|---|---|---|---|
| CS511.1 | Course | Thursday 17:00 - 19:50 | B F1.16 | - | - |
Office Hours & Room
| Day | Time | Office | Notes |
|---|---|---|---|
| Monday | 11:00 - 15:00 | A F2.6 | |
| Thursday | 11:00 - 15:00 | A F2.6 |
Assessment Methods and Criteria
Assessment Components
Final Project: research paper and presentation
AI: Consult InstructorAlignment with Learning Outcomes : 1 2 3 4 5
Lab assignments (weekly notebooks)
AI: Consult InstructorAlignment with Learning Outcomes : 1 2 3 4
Research paper presentation and discussion
AI: Consult InstructorAlignment with Learning Outcomes : 5
Research project proposal (Week 8)
AI: Consult InstructorAlignment with Learning Outcomes : 5
Research project progress report (Week 12)
AI: Consult InstructorAlignment with Learning Outcomes : 5
Participation in paper discussions
AI: Not AllowedAlignment with Learning Outcomes : 5
IUS Grading System
| Grading Scale | IUS Grading System | IUS Coeff. | Letter (B&H) | Numerical (B&H) |
|---|---|---|---|---|
| 0 - 44 | F | 0 | F | 5 |
| 45 - 54 | E | 1 | ||
| 55 - 64 | C | 2 | E | 6 |
| 65 - 69 | C+ | 2.3 | D | 7 |
| 70 -74 | B- | 2.7 | ||
| 75 - 79 | B | 3 | C | 8 |
| 80 - 84 | B+ | 3.3 | ||
| 85 - 94 | A- | 3.7 | B | 9 |
| 95 - 100 | A | 4 | A | 10 |
Late Work Policy
Information about late submission policies will be shared during class and posted in this section. Please check back for official guidelines.
ECTS Credit Calculation
📚 Student Workload
This 6 ECTS credit course corresponds to 150 hours of total student workload, distributed as follows:
Lectures and labs (in class)
45 hours ⏳ (15 week × 3 h)
Reading assigned papers and lecture review
28 hours ⏳ (14 week × 2 h)
Completing lab notebooks and exercises
26 hours ⏳ (13 week × 2 h)
Preparation of the paper presentation
6 hours ⏳ (1 week × 6 h)
Research project (proposal, experiments, paper, talk)
45 hours ⏳ (15 week × 3 h)
150 Total Workload Hours
6 ECTS Credits
Course Policies
Academic Integrity
All work submitted must be your own. Plagiarism, cheating, or any form of academic dishonesty will result in disciplinary action according to university policies. When in doubt about citation practices, consult the instructor.
Attendance Policy
Students are expected to adhere to the attendance requirements as outlined in the International University of Sarajevo Study Rules and Regulations. Excessive absences, whether excused or unexcused, may impact academic performance and eligibility for assessment. Mandatory sessions (e.g., labs, workshops) require attendance unless formally exempted. For detailed policies on absences, documentation, and penalties, please refer to the official university regulations.
Technology & AI Policy
Laptops/tablets may be used for note-taking only during lectures. Phones should be silenced and put away during all class sessions. Audio/video recording requires prior permission from the instructor.
Artificial Intelligence (AI) Usage: The use of AI tools (e.g., ChatGPT, Copilot, Gemini) varies by assessment component. Please refer to the AI usage indicator next to each assessment item in the Assessment Methods and Criteria section above. Submitting AI-generated content as your own work, where AI is not explicitly allowed, constitutes an academic integrity violation.
Communication Policy
All course-related communication should occur through official university channels (institutional email or SIS). Emails should include [CS511] in the subject line.
Academic Quality Assurance Policy
Course Academic Quality Assurance is achieved through Semester Student Survey. At the end of each academic year, the institution of higher education is obliged to evaluate work of the academic staff, or the success of realization of the curricula.
Learning Tips
Be prepared to contribute thoughtfully during class discussions, labs, or collaborative work. Active participation deepens understanding and encourages critical thinking.
Complete assigned readings or prep materials before class. Take notes, highlight key ideas, and jot down questions. Aim to grasp core concepts and their applications—not just facts.
Use course frameworks or methodologies to analyze problems, case studies, or projects. Begin early to allow time for reflection and refinement. Seek feedback to improve your work.
Don’t hesitate to reach out when something is unclear. Use office hours, discussion boards, or peer networks to clarify concepts and stay on track.
Syllabus Last Updated on Oct 10, 2026 | International University of Sarajevo
Print Syllabus
Referencing Curricula Print this page
| Course Code | Course Title | Weekly Hours* | ECTS | Weekly Class Schedule | ||||||
| T | P | |||||||||
| CS511 | Advanced Artificial Intelligence | 3 | 0 | 6 | ||||||
| Prerequisite | None | It is a prerequisite to | - | |||||||
| Lecturer | Ali Almisreb | Office Hours / Room / Phone | Monday: 11:00-15:00 Thursday: 11:00-15:00 |
|||||||
| aalmisreb@ius.edu.ba | ||||||||||
| Assistant | Assistant E-mail | |||||||||
| Course Objectives | This graduate course starts where AID401 Deep Learning ends (the Transformer, CNNs, RL basics) and studies the three pillars of current AI research in technical depth: large language models (LLMs), vision-language models (VLMs) and LLM-based agents. Every topic is developed with equations, derivations and worked numeric examples, implemented in weekly PyTorch / Hugging Face labs on open-weight models, and connected to recent research papers. Students present a research paper and carry out a semester-long, reproducible research project. Learning outcomes. On successful completion, students will be able to: Derive and analyse modern LLM architectures and pretraining: tokenization, RoPE, GQA, scaling laws, distributed training (ZeRO/FSDP), FlashAttention, long context and mixture-of-experts. Implement and evaluate post-training and inference methods: SFT and LoRA/QLoRA, reward models, RLHF/PPO and DPO, reasoning with GRPO and test-time compute, quantization and speculative decoding. Explain, build and evaluate vision-language models: contrastive pretraining (CLIP/SigLIP), connectors, grounding, hallucination, and unified multimodal generation with discrete visual tokens. Design, build and evaluate LLM agents with retrieval (RAG), memory and tools (function calling, MCP), and analyse their reliability and security (prompt injection, adversarial attacks). Critically read current research and conduct a reproducible research project with sound experimental design and statistical reporting, presented as a paper and a talk. |
|||||||||
| Textbook | Jurafsky, D., & Martin, J. H. Speech and Language Processing (3rd ed. draft, 2025). Online: https://web.stanford.edu/~jurafsky/slp3/ Raschka, S. Build a Large Language Model (From Scratch). Manning, 2024. Prince, S. J. D. Understanding Deep Learning. MIT Press, 2023. Online: https://udlbook.github.io/udlbook/ Bishop, C. M., & Bishop, H. Deep Learning: Foundations and Concepts. Springer, 2024. | |||||||||
| Additional Literature |
|
|||||||||
| Learning Outcomes | After successful completion of the course, the student will be able to: | |||||||||
|
||||||||||
| Teaching Methods | Each week combines a technical lecture with a hands-on lab: concepts are first derived mathematically and checked on small numeric examples, then implemented from scratch and with standard libraries, and finally evaluated on real models and benchmarks. Students read the assigned papers before class, each student presents and leads the discussion of one recent research paper, and teams of 1–3 students carry out a semester-long research project on LLMs, VLMs or agents, reported as a workshop-style paper with a reproducibility package. | |||||||||
| Teaching Method Delivery | Face-to-face | Teaching Method Delivery Notes | ||||||||
| WEEK | TOPIC | REFERENCE | ||||||||
| Week 1 | Modern LLM Architecture and Scaling | Vaswani et al. (2017); Hoffmann et al. (2022); Raschka (2024) Ch. 2–4 | ||||||||
| Week 2 | Pretraining at Scale: Data and Distributed Training | Rajbhandari et al. (2020) ZeRO; Shoeybi et al. (2019) Megatron-LM | ||||||||
| Week 3 | Long Context, Fast Attention and Mixture of Experts | Dao et al. (2022) FlashAttention; Peng et al. (2024) YaRN; Fedus et al. (2022) Switch Transformer | ||||||||
| Week 4 | Post-training I: SFT and Parameter-Efficient Fine-Tuning | Hu et al. (2022) LoRA; Dettmers et al. (2023) QLoRA; Zhou et al. (2023) LIMA | ||||||||
| Week 5 | Post-training II: RLHF and DPO | Ouyang et al. (2022) InstructGPT; Schulman et al. (2017) PPO; Rafailov et al. (2023) DPO | ||||||||
| Week 6 | Reasoning and Test-Time Compute | Wei et al. (2022); Wang et al. (2023); Shao et al. (2024) GRPO; DeepSeek-AI (2025) DeepSeek-R1 | ||||||||
| Week 7 | LLM Inference and Serving | Kwon et al. (2023) vLLM; Frantar et al. (2023) GPTQ; Leviathan et al. (2023) | ||||||||
| Week 8 | Midterm Week: LLM Review and Research Proposal Presentations | Lecture notes and labs of Weeks 1–7 | ||||||||
| Week 9 | Vision-Language Model Architectures | Radford et al. (2021) CLIP; Zhai et al. (2023) SigLIP; Liu et al. (2023) LLaVA; Wang et al. (2024) Qwen2-VL | ||||||||
| Week 10 | Grounding, Perception and VLM Evaluation | Minderer et al. (2023) OWLv2; Li et al. (2023) POPE; Leng et al. (2024) VCD | ||||||||
| Week 11 | Unified Multimodal Models | van den Oord et al. (2017) VQ-VAE; Chang et al. (2022) MaskGIT; Mentzer et al. (2024) FSQ; Zhou et al. (2024) Transfusion | ||||||||
| Week 12 | Retrieval and Memory | Karpukhin et al. (2020) DPR; Lewis et al. (2020) RAG; Khattab & Zaharia (2020) ColBERT | ||||||||
| Week 13 | LLM and Multimodal Agents | Yao et al. (2023) ReAct; Schick et al. (2023) Toolformer; Anthropic (2024) MCP specification | ||||||||
| Week 14 | Training, Evaluating and Securing Agents | Yao et al. (2024) tau-bench; Greshake et al. (2023); Zou et al. (2023) GCG; Debenedetti et al. (2025) CaMeL | ||||||||
| Week 15 | Course Synthesis and Final Project Presentations | Lecture notes and labs of Weeks 9–14 | ||||||||
| Assessment Methods and Criteria | Evaluation Tool | Quantity | Weight | Alignment with LOs | AI Usage |
| Final Project: research paper and presentation | 1 | 40 | 1,2,3,4,5 | Consult Instructor | |
| Semester Evaluation Components | |||||
| Lab assignments (weekly notebooks) | 13 | 20 | 1,2,3,4 | Consult Instructor | |
| Research paper presentation and discussion | 1 | 15 | 5 | Consult Instructor | |
| Research project proposal (Week 8) | 1 | 10 | 5 | Consult Instructor | |
| Research project progress report (Week 12) | 1 | 10 | 5 | Consult Instructor | |
| Participation in paper discussions | 14 | 5 | 5 | Not Allowed | |
| *** ECTS Credit Calculation *** | |||||
| Activity | Hours | Weeks | Student Workload Hours | Activity | Hours | Weeks | Student Workload Hours | |||
| Lectures and labs (in class) | 3 | 15 | 45 | Reading assigned papers and lecture review | 2 | 14 | 28 | |||
| Completing lab notebooks and exercises | 2 | 13 | 26 | Preparation of the paper presentation | 6 | 1 | 6 | |||
| Research project (proposal, experiments, paper, talk) | 3 | 15 | 45 | |||||||
| Total Workload Hours = | 150 | |||||||||
| *T= Teaching, P= Practice | ECTS Credit = | 6 | ||||||||
| Course Academic Quality Assurance: Semester Student Survey | Last Update Date: 06/10/2026 | |||||||||