MOD-01
FOUNDATION
Python for Data Science
Memory models, list comprehensions, NumPy vectorization, broadcast mechanics, and structured arrays.
Read SyllabusINNOVATE. IMPLEMENT. INSPIRE.
A student-driven technical community building, learning and experimenting with AI and machine learning.
ORIENTAL COLLEGE OF TECHNOLOGY
•
BHOPAL · INDIA
[23.2599° N, 77.4126° E]
LEARN
→
BUILD
→
RESEARCH
→
EXPERIMENT
→
SHIP
→
SHARE
METRIC 01
12
Curriculum Modules
From NumPy & Calculus to LLM Agents & MLOps
METRIC 02
04
Colab Notebook Labs
Zero-setup executable cloud GPU environments
METRIC 03
09+
Verified Events
Bootcamps, Aptify 2.0, CSC, Hackathons
METRIC 04
41
Council & Team
Faculty coordinators, architects, wing leads
00 // EDITORIAL IDENTITY & MANIFESTO
We Don't Just Study AI.
We Build with It.
Surface-level tutorials cannot prepare engineers for production reality. We enforce mathematical foundations, zero-dependency algorithm rebuilds, strict peer review, and continuous deployment.
PHASE 01 // FOUNDATION
Mathematical Rigor
Mastering matrix calculus, eigenvalues, probability distributions, gradient descent convergence, and loss function geometry before invoking high-level frameworks.
NumPy • SymPy • SciPy
PHASE 02 // RECONSTRUCTION
Zero-Dependency Rebuilds
Implementing fundamental algorithms from scratch—linear regression, k-nearest neighbors, naive Bayes, and multilayer perceptrons—using pure Python and analytical calculus.
Pure Python 3.12+ • Pure Math
PHASE 03 // DEEP LEARNING
Production Frameworks
Accelerating architectures with PyTorch tensors, autograd, GPU acceleration, and Scikit-Learn pipelines. Training reproducible models with strict cross-validation.
PyTorch • Scikit-Learn • CUDA
PHASE 04 // SERVING
Inference & Optimization
Packaging model weights into ONNX runtimes, building sub-20ms FastAPI microservices, implementing token streaming, and containerizing pipelines for Kubernetes deployment.
FastAPI • Docker • ONNX Runtime
PHASE 05 // RIGOR
Peer Review & PR Audit
Every student pull request undergoes automated linting, test coverage checks, and rigorous review by club technical leads before merging into production.
GitHub Actions • Ruff • Pytest
PHASE 06 // ECOSYSTEM
Open Source Impact
Publishing benchmark studies, dataset releases, Colab tutorials, and reusable starters to empower the next generation of engineers across Central India.
MIT License • Hugging Face Hub
01 // LEARN
WE LEARN THE FOUNDATIONS.
Curated syllabus spanning foundational Python to advanced frontier models. Every module includes theory notebooks, exercises, and evaluation criteria hosted in our open repository.
MOD-02
MATHEMATICS
Mathematics for Machine Learning
Linear algebra (SVD, PCA), multivariate calculus (Jacobians, Hessians), probability distributions, and Bayes theorem.
Read Syllabus
MOD-03
DATA OPS
Data Wrangling & Analysis
Pandas & Polars, missing value imputation strategies, categorical encoding, Outlier bounds, and Seaborn visualization.
Read Syllabus
MOD-04
CLASSICAL ML
Classical Machine Learning
Ordinary Least Squares, Ridge/Lasso, Logistic regression, SVMs, Decision Trees, Random Forests, and XGBoost.
Read Syllabus
MOD-05
DEEP LEARNING
Deep Learning Fundamentals
Multilayer perceptrons, backpropagation equations, activation functions, optimizers (AdamW, SGD with momentum), and PyTorch tensors.
Read Syllabus
MOD-06
VISION
Computer Vision & CNNs
Convolutions, pooling, ResNet residual blocks, transfer learning, object detection with YOLO, and real-time OpenCV streams.
Read Syllabus
MOD-07
NLP
Natural Language Processing
Tokenization, TF-IDF vectors, Word2Vec embeddings, sequence modeling with LSTMs, and sentiment classification.
Read Syllabus
MOD-08
TRANSFORMERS
Transformers & LLMs
Scaled dot-product self-attention, multi-head attention, positional encodings, BERT, GPT decoder architectures, and quantization.
Read Syllabus
MOD-09
GENAI
Generative AI & RAG
Dense vector embeddings, pgvector, hybrid BM25 + dense search, semantic chunking, and verifiable grounded prompt generation.
Read Syllabus
MOD-10
MLOPS
MLOps & Production Systems
Model serialization, FastAPI inference endpoints, Docker multi-stage builds, model drift monitoring, and batch processing.
Read Syllabus
MOD-11
RL & AGENTS
Reinforcement Learning
Markov Decision Processes, Bellman optimality, Q-learning, Policy Gradients, and tool-augmented autonomous LLM agents.
Read Syllabus
MOD-12
RESEARCH
Research Methodology & Ethics
Paper replication standards, algorithmic fairness auditing, bias mitigation, ablation studies, and IEEE/arXiv paper authoring.
Read Syllabus
EXECUTABLE COLAB SANDBOXES
Interactive Cloud GPU Notebooks
QUICK CLONE:
git clone https://github.com/AIMLCLUBOCT/Workshops.git
Python, NumPy & Pandas Primer
Fundamental vector operations, memory allocation benchmarks, array broadcasting, and clean data wrangling on real academic datasets.
Vectorized arithmetic
Pandas indexing speedups
Hands-on exercises
Supervised Machine Learning
Linear regression, logistic classifiers, decision boundaries, ROC-AUC curves, and cross-validation grids using Scikit-Learn.
Confusion matrix analysis
Hyperparameter tuning
Overfitting prevention
PyTorch Neural Networks from Scratch
Writing custom `nn.Module` classes, understanding automatic differentiation via `loss.backward()`, learning rate schedulers, and image classification.
Dynamic computation graphs
DataLoader & batching
Loss monitoring curves
Sentiment & Text NLP Pipeline
Text normalization, subword tokenizers, TF-IDF matrix transformations, Naive Bayes vs. pre-trained HuggingFace transformers.
Text cleaning regular expressions
Precision / Recall curves
Inference benchmark
02 // BUILD
FROM IDEAS TO SYSTEMS.
Production-grade systems created by OCT engineers. Every project includes automated test coverage, modular pipelines, and zero-dependency mathematical fallbacks.
RAW INPUT STREAM
Email MIME Body / RFC 5322 Headers
PREPROCESSING ENGINE
Regex Normalization • TF-IDF Vectorizer
CLASSIFICATION CORE
MultinomialNB • Fallback Vector Math
OUTPUT ENVELOPE
Spam Probability • Sub-8ms Latency
ACCURACY
97.4%
LATENCY
< 8ms
COVERAGE
100%
Campus Phishing Email Detection System
The Threat: Collegiate networks frequently suffer credential harvesting attacks targeting university student portals and fee gateways.
The Solution: A dual-mode text classifier using TF-IDF tokenization combined with Multinomial Naive Bayes. Includes a pure Python zero-dependency mathematical fallback module for offline embedded devices.
Scikit-Learn
FastAPI
Pure Python Math
Pytest
Inspect Architecture
Dual-Mode Fallback Enabled
Student Academic Performance & Risk Predictor
The Problem: Identifying at-risk engineering undergraduates before end-of-semester university examinations allows early academic intervention.
The Solution: A tabular regression and classification model analyzing internal assessment scores, laboratory attendance, and study hours. Features an analytical gradient descent fallback engine.
Random Forest
Gradient Descent
Pandas
CSV Pipelines
Inspect Architecture
R² Score: 0.91
Construction Safety Helmet & PPE Real-Time Vision
The Problem: Industrial manufacturing and campus construction sites require strict personal protective equipment compliance to prevent injuries.
The Solution: Fine-tuned YOLOv8 neural network trained on hard-hat and vest datasets, optimized with ONNX Runtime to maintain 45 FPS on edge video streams.
YOLOv8
OpenCV
ONNX Runtime
PyTorch
Inspect Architecture
Edge Speed: 45 FPS
Campus Ordinance Semantic RAG Assistant
The Problem: University syllabi, exam rules, and college notifications span hundreds of unstructured PDF pages, making student inquiry tedious.
The Solution: Hybrid BM25 keyword + dense vector semantic retrieval system over official OCT documents with citation tracking and hallucination guardrails.
LangChain
pgvector
FastAPI
SentenceTransformers
Inspect Architecture
Zero Hallucination Gate
03 // RESEARCH
ASK. EXPERIMENT. MEASURE.
Moving beyond theoretical speculation. Our student researchers conduct empirical replications, model quantizations, and inference efficiency audits.
EXP-01 // QUANTIZATION
OCT-LAB 2026
Edge Latency vs INT8 Quantization Loss
Quantizing ResNet-50 and YOLOv8 weights from FP32 to INT8 via ONNX Runtime to evaluate mAP degradation against inference frame rate gains on low-power Intel Celeron and Raspberry Pi edge devices.
PRECISION
FPS (CPU)
mAP@50
SIZE
FP32 Original
12.4
88.2%
98 MB
INT8 Quantized
38.7 (3.1x)
86.9% (-1.3%)
26 MB (-73%)
EXP-02 // VISION ROBUSTNESS
OCT-LAB 2025
MediaPipe Landmark Occlusion in Low-Light Classrooms
Evaluating 33-point pose landmark estimation drift under fluctuating ambient illumination (50–300 lux) and classroom desk occlusions for student posture analysis.
CONDITION
CONFIDENCE
DRIFT (PX)
FPS
High Light (300 lux)
0.94
± 2.1 px
60 FPS
Low Light (50 lux)
0.81
± 6.8 px
58 FPS
EXP-03 // NLP PARAMETERS
OCT-LAB 2026
PEFT / LoRA vs Full Fine-Tuning Convergence
Comparing parameter-efficient fine-tuning (LoRA rank=8, alpha=16) against full backpropagation on Llama-3-8B for campus ordinance question answering on 1x 16GB GPU.
METHOD
TRAINABLE PARAMS
VRAM USE
ROUGE-L
Full Tuning
8.03 Billion
OOM (>48GB)
N/A
LoRA (r=8)
13.6 Million (0.17%)
11.4 GB
0.724
DEEP LEARNING FOUNDATIONS
Scaled Dot-Product Attention & Multi-Head Lattice Mechanics
Students dissect the mathematical core of the Transformer architecture: query-key-value projections Attention(Q, K, V) = softmax(QK^T / sqrt(d_k))V, causal masking matrices, rotary position embeddings (RoPE), and KV-caching memory profiles for real-time edge execution.
O(N²) → O(N)
Complexity Bounds
12 Heads
Parallel Projections
768 Dim
Latent Space
04 LIVE ACTIVITIES & NOTICES
Stay synchronized with club operations. Live activities, workshops, and releases.
Whenever an organizer schedules a workshop, a student commits an open-source project blueprint, or new learning materials are published, it appears right here in real-time. Students can also propose their own activities directly.
05 // CHRONOLOGY
VERIFIED EVENT ARCHIVE.
From technical aptitude challenges to deep-learning bootcamps. Historical records of hands-on sessions conducted at Oriental College of Technology, Bhopal.
2026-02-15
COMPETITION & AWARDS
Aptify 2.0 & Grand Winners Felicitation Ceremony
The flagship second edition of Aptify featuring advanced technical aptitude, algorithmic reasoning rounds, and a grand felicitation ceremony honoring campus winners with certificates and accolades.
2026-01-28
CONVENTION
CSC 2026 Technical Convention & AI Showcase
Annual technical convention showcasing student innovations, model demonstration booths, and research poster presentations across AI & Machine Learning.
2025-12-30
ORIENTATION
New Core Team Member Orientation
Official onboarding session for newly inducted executive council and wing leads. Presentation of 2026 development roadmap, Git workflows, and event calendars.
2025-12-13
WORKSHOP
WordPress & Web Architecture Workshop
Hands-on session introducing modern web architecture, CMS deployment, DNS management, and custom endpoint integrations for college technical societies.
2025-11-29
EXPERT TALK
Expert Seminar: Reinforcement Learning & MDPs
Deep-dive into Markov Decision Processes, policy gradient theorem, reward engineering, and autonomous robotics applications presented by senior industry practitioners.
2025-11-10
COMPETITION
Codify — Flagship Algorithmic Coding Challenge
Intra-college competitive programming tournament testing algorithmic efficiency, dynamic programming, tree traversals, and Python mathematical problem-solving.
2025-10-13
COMPETITION
Apfity — Aptitude & Logical Problem Solving
Comprehensive challenge assessing numerical aptitude, pattern recognition, and computational thinking for engineering pre-final and final year candidates.
2025-09-12
BOOTCAMP
3-Day Hands-on AI/ML Bootcamp (10–12 Sep Workshop)
Intensive 3-day practical bootcamp covering Python fundamentals, NumPy vectorization, exploratory data analysis, and predictive modeling for engineering students.
2025-09-11
WORKSHOP
DSPL 2-Day Intensive Hands-On Workshop
Data Structures & Programming Logic bootcamp focusing on recursion, pointer mechanics, stack/queue paradigms, and practical Python implementations.
2025-09-10
SESSION
DSPL Foundation Session: Logic & Algorithmic Thinking
Introductory symposium laying the foundation for mathematical abstraction and clean procedural design in modern high-level languages.
2025-08-26
EXPERT TALK
Tech Industry Practices with Coding Thinker
Interactive dialogue on corporate engineering standards, clean architecture, software engineering hiring patterns, and technical resume development.
2025-08-14
INAUGURATION
Official AIML Club Inauguration & Faculty Keynote
Foundational assembly inaugurating the club charter, unveiling the student council, and establishing research partnerships with Oriental Group leadership.
06 // MEDIA & EVIDENCE
AUTHENTIC PHOTOGRAPHIC ARCHIVES.
High-resolution visual evidence of campus operations, student award felicitations, competitive programming hackathons, and laboratory environments at OCT Bhopal. Click any photograph to inspect full-screen.
CAMPUS ARCHITECTURE
Oriental College of Technology Campus Center
The flagship campus facility housing computer science laboratories, high-compute setups, and seminar halls.
COMPETITION & AWARDS
Aptify 2.0 Grand Winners Felicitation Ceremony
Student winners honored on stage with technical trophies, certificates, and recognition by the core council.
CAMPUS LAB EVALUATION
Proctored Algorithmic Aptitude & Coding Round
Engineering students actively competing in computational logic, pattern recognition, and live problem-solving.
STUDENT BOOTCAMP
3-Day Hands-on AI/ML Practical Bootcamp
Intensive student lab sessions covering NumPy vectorization, exploratory data analysis, and machine learning pipelines.
FOUNDING MILESTONE
Official AIML Club Inauguration Assembly
Foundational assembly inaugurating the club charter, induction of the core team, and faculty keynote.
CONVENTION & EXHIBITION
CSC 2026 Technical Convention & Showcase
Student technical symposium, applied machine learning project demonstrations, and campus tech exhibits.
[VERIFIED GOOGLE DRIVE ARCHIVE]
Open Full Media Drive Archive
21 Dedicated Event Folders & Certificate Archives
Access original high-resolution photo galleries from the 10–12 Sep DSPL Bootcamp, Codify sprints, SnapCode rounds, and founding inauguration albums directly on Google Drive.
07 // HUMAN CAPITAL
THE 41-MEMBER COUNCIL.
Meet the faculty coordinators, executive leadership, and wing architects driving daily research, operations, and workshops at Oriental College of Technology.
08 // OPEN SOURCE
BUILD IN PUBLIC. SHIP TO PRODUCTION.
All curriculum roadmaps, Colab labs, production microservice blueprints, and event logs are openly licensed under MIT across the AIMLCLUBOCT organization.
The primary 12-module open-source curriculum spanning NumPy, Calculus, Scikit-Learn, PyTorch, Transformers, and MLOps.
Hands-on Google Colab notebooks tailored for live student bootcamps with 1-click cloud execution.
Curated starter projects, fallbacks, and production-grade architectures with good first issues for contributors.
Standardized blueprints for ML microservices, Docker inference containers, evaluation benchmarks, and issues.
Centralized archive of all club hackathons, workshops, bootcamps (Aptify 2.0, Codify, SnapCode) with schedules and outcomes.
Organization-wide governance, community health profiles, constitution, code of conduct, and contribution workflows.
[THE CHARTER]
BUILD WHAT
COMES NEXT.
Artificial intelligence is not a passive spectator sport. Whether you are in your first semester writing your first matrix transpose or a final-year researcher training sparse attention decoders—our laboratory is open.
Club Media & Event Records
Official Google Drive repository archiving high-resolution event photographs, certificates (10–12 Sep), and Aptify 2.0 ceremony moments.
drive.google.com/drive/folders/1-_byssQs...
Open Media Archive
Flagship Production Web
Next.js production web portal hosting real-time notifications, event registrations, and announcements.
aimlcluboct.in
Launch Site
Verified WhatsApp Channel
Direct announcement broadcast for competition alerts, workshop deadlines, and guest speaker sessions.
WhatsApp Broadcast Channel
Follow Channel
Commudle Community
Official community partner platform for event ticketing, campus check-ins, and hackathon certificate issuance.
commudle.com/communities/ai-ml-club
Join Commudle