Skip to content
View nabeegh-khan's full-sized avatar
  • University of Toronto
  • Toronto, ON

Block or report nabeegh-khan

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
nabeegh-khan/README.md

Nabeegh Khan

MEng candidate in Electrical & Computer Engineering at the University of Toronto, graduating December 2026, with an emphasis in Data Analytics and Machine Learning. P.Eng, PMP, and around six years of professional engineering experience.

I work on machine learning for wireless communications, on retrieval-augmented and NLP systems, and on end-to-end ML pipelines, alongside engineering-education research as a graduate research assistant. Recent projects span parameter-efficient fine-tuning of a wireless foundation model, reinforcement and supervised learning for MIMO power allocation, beam prediction on real vehicle-to-vehicle measurements, a retrieval-augmented QA system over 3GPP standards with source citations, mixed-methods and bibliometric studies of generative AI in education, and a streaming MLOps pipeline. I try to build things that are reproducible and reported honestly, including the results that came out modest or negative.

Selected projects

Project What it does Stack
LWM-LoRA: Scenario-Adaptive mmWave Beam Prediction LoRA fine-tuning of the Large Wireless Model (LWM v1.1) for 64-beam mmWave prediction across three DeepMIMO scenarios. Rank-4 adapters train 4.82% of parameters and reach 76.8% top-1 on Miami; cross-scenario transfer with 20% of target data matches full fine-tuning within 0.3%; ONNX INT8 export cuts model size 69.5%. PyTorch, HuggingFace, LoRA/PEFT, DeepMIMOv3, ONNX Runtime, W&B
3GPP Specification Assistant (RAG) Retrieval-augmented QA over 14 3GPP NR specs (4,493 pages, 18,187 chunks). Answers come back with source citations, evaluated with RAGAS (faithfulness 0.675, context recall 0.750). LangChain, ChromaDB, GPT-4o-mini, FastAPI, Streamlit, RAGAS
DeepSense 6G V2V Beam Prediction mmWave beam prediction on 112,189 real V2V measurements (Scenarios 36–39, 60 GHz). Random Forest reached 22.6% top-1, beating every deep model I tried; a DQN analysis traced its ceiling to feature compression. PyTorch, scikit-learn, DeepSense 6G, Gymnasium
AI in the Classroom: Mixed-Methods Survey + Reddit Convergent mixed-methods study integrating survey data (n=625) with 465 Reddit posts. Ordinal regression found attitude toward use the dominant predictor of AI adoption (OR=8.32, p<0.001). BERTopic, VADER, spaCy, statsmodels, Reddit API
6G Massive MIMO Resource Allocation DQN versus supervised learning for power allocation in a 7-cell, 70-user Massive MIMO environment. Supervised reward regression matched the DQN controller at 4.4× the random baseline, a cheaper alternative to RL on this problem. PyTorch, Stable-Baselines3, Gymnasium
Real-Time Anomaly Detection MLOps Streaming anomaly-detection pipeline on the NAB benchmark: Kafka to Spark Structured Streaming to a DuckDB feature store, an LSTM autoencoder tracked in MLflow, FastAPI serving, Airflow orchestration, Evidently drift monitoring. ROC-AUC 0.64 on a hard unsupervised task, reported with its limitations. PyTorch, Kafka, Spark, MLflow, FastAPI, Airflow, Evidently, DuckDB
AI-in-Education Bibliometric + NLP Analysis Collected 4,403 papers via the OpenAlex and Semantic Scholar APIs; BERTopic surfaced 27 research clusters; a chi-square test confirmed a post-ChatGPT topic shift (χ²=323.87, p<0.0001). BERTopic, VADER, OpenAlex API, pandas, scipy
Ontario Electricity Demand Forecasting Hourly demand forecasting on 109,000+ records from IESO, Environment Canada, and NASA POWER, with 13 engineered features. A 3-layer network cut RMSE 50.7% over baseline (R²=0.993). scikit-learn, XGBoost, PyTorch, pandas
Big Data Analytics: Spark & Azure Synapse Distributed processing with Spark (RDD and DataFrame APIs, Scala on Databricks) and cloud SQL analytics on Azure Synapse over multi-file corpora and partitioned retail data. Apache Spark, Scala, Databricks, Azure Synapse, T-SQL

Skills

Machine learning: PyTorch, scikit-learn, deep learning, CNN/LSTM/RNN, reinforcement learning (DQN, Stable-Baselines3), Random Forest, SVM, feature engineering

Foundation models & transfer learning: HuggingFace Transformers, LoRA/PEFT, Large Wireless Model, transfer learning, ONNX Runtime, INT8 quantization, Weights & Biases

MLOps & serving: MLflow, Airflow, Evidently, Docker, FastAPI, model serving, drift monitoring

Streaming & data engineering: Kafka, Spark Structured Streaming, DuckDB, dbt, Databricks, Azure Synapse

LLM & RAG: LangChain, ChromaDB, OpenAI embeddings, GPT-4o-mini, RAGAS, LangSmith, Streamlit

NLP & text analytics: BERTopic, VADER, spaCy, topic modeling, qualitative coding

Statistics & research methods: chi-square, ANOVA, Mann-Whitney U, ordinal logistic regression, mixed-methods research, hypothesis testing, statsmodels, scipy

Wireless: mmWave beam prediction, beamforming, Massive MIMO, V2V, DeepMIMOv3, DeepSense 6G, Gymnasium

Languages & tools: Python, Scala, SQL, Git, Jupyter, pandas, NumPy, Matplotlib, Seaborn, Plotly

Now

  • Research Assistant at ISTEP, University of Toronto, currently and in summer 2025. Quantitative and qualitative analysis of engineering-education survey data using chi-square, Mann-Whitney U, and mixed-methods approaches; co-authored and presented a poster at UTERC 2025.
  • Finishing my MEng in Electrical & Computer Engineering, graduating December 2026.

A note on tooling

These projects were built with significant AI-assisted coding; I used Claude (Anthropic) as a coding assistant. I scoped the questions, chose the datasets and methods, and ran, validated, and interpreted the results. Each repository documents this in its README.

Pinned Loading

  1. 6g-lwm-beam-prediction 6g-lwm-beam-prediction Public

    LoRA fine-tuning of the LWM wireless foundation model for 64-beam mmWave prediction across three DeepMIMO scenarios. Rank-4 adapters (4.82% of parameters) match full fine-tuning within 0.3 points o…

    Jupyter Notebook

  2. deepsense-6g-beam-prediction deepsense-6g-beam-prediction Public

    Real-world mmWave beam prediction on 112,189 DeepSense 6G V2V measurements. Compares CNN, LSTM, RNN, SVM, Random Forest, and DQN; Random Forest beats every deep model at 22.6% top-1.

    Jupyter Notebook 1 1

  3. 6g-mimo-resource-allocation 6g-mimo-resource-allocation Public

    DQN vs supervised learning for power allocation in a 7-cell, 70-user Massive MIMO environment. Supervised reward regression matched the DQN at 4.4× the random baseline, using only 3,000 samples.

    Jupyter Notebook

  4. 3gpp-rag 3gpp-rag Public

    Retrieval-augmented QA over 14 3GPP NR specs (4,493 pages), with cited answers and RAGAS evaluation. Ask a spec question, get an answer grounded in the source with the page cited.

    Jupyter Notebook 2

  5. real-time-anomaly-mlops real-time-anomaly-mlops Public

    End-to-end streaming MLOps pipeline for time-series anomaly detection on the NAB benchmark: Kafka, Spark, MLflow, FastAPI, Airflow, and Evidently wired together across the full lifecycle.

    Jupyter Notebook

  6. ai-education-mixed-methods ai-education-mixed-methods Public

    Mixed-methods study of student AI perceptions, triangulating survey data (n=625) with 465 Reddit posts. Surveys rate AI highly while the academic-writing discourse is the most negative topic, a uti…

    Jupyter Notebook