Machine Learning Engineer  ·  Georgia Tech  ·  NUST

HUZAIFAH WASIM

Machine learning engineer — generative AI, reinforcement learning, and robotics.

0 GPA at Georgia Tech
0+ Projects
0 Publications

Selected Work

Showcase

The Full Archive

All projects.

Smart Garbage Bin

Edge-deployed waste classifier: MobileNet v2 retrained on custom garbage data, integrated with a deep RL agent, and explained with LIME, all running on Jetson Nano.

  • Jetson Nano
  • TensorFlow
  • MobileNet v2
  • OpenAI Gym
  • +1

Scalable RISC-V Multicore Processor

Custom 32-bit RISC-V core integrated into the OpenPiton multicore framework with set-associative caches and AXI-4 interconnect, deployed on a Genesys2 FPGA.

  • Xilinx Vivado
  • OpenPiton
  • FPGA
  • RISC-V
  • +1

APF + DQN for TurtleBot3 Navigation

Seeded a DQN with an Artificial Potential Field controller to cut training time for autonomous robot navigation in Gazebo.

  • ROS
  • TensorFlow
  • Python
  • Gazebo

SLAM + DQN for TurtleBot3 Navigation

Fused simultaneous localization and mapping with a Deep Q-Network to give a TurtleBot3 both spatial awareness and learned navigation policy.

  • ROS
  • TensorFlow
  • SLAM
  • Deep Q-Network
  • +1

Fine-tuned French GPT-3 Chatbot

A web app for fine-tuned French language generation with GPT-3, letting users retrain the model on their own datasets.

  • OpenAI API
  • GPT-3
  • Flask
  • Python

Predictive Object Tracking

Kalman filter-based object tracker that predicts future positions in real time, built as a navigation primitive for autonomous vehicles.

  • Python
  • Computer Vision
  • Kalman Filter
  • State Estimation

Speech Recognition on Edge

Hybrid HMM/GMM speech recognition model built for edge deployment on FPGAs, with no cloud dependency.

  • Kaldi
  • HMM
  • GMM
  • FPGA
  • +1

Sign Language Translation with OAK-D

Real-time sign language translation using the OAK-D depth camera, a finalist in the OpenCV AI Competition 2021.

  • OpenCV
  • OAK-D
  • DepthAI
  • Computer Vision

Robot Localization with Particle Filters

Particle filter with a custom fuzzing algorithm for autonomous vehicle localization across uncertain map conditions.

  • Python
  • Particle Filters
  • Probabilistic Inference

Robot Path Planning

A* search and Value Iteration implemented for autonomous warehouse robot navigation.

  • Python
  • A* Search
  • Value Iteration
  • Dynamic Programming

Short-term Residential Load Forecasting

Benchmarked Transformers, LSTMs, and RL agents for household energy consumption forecasting, comparing architectures head-to-head on real load data.

  • TensorFlow
  • Transformers
  • LSTM
  • Deep Q-Network

Quadcopter PID Controller

Designed and tuned a PID controller for stable quadcopter flight.

  • Python
  • Control Systems
  • PID
  • Robotics

Task Scheduling in Cloud Computing

Topological sort-based scheduler for cloud computing workflows that stages and executes dependent tasks across distributed pipelines.

  • Python
  • Topological Sort
  • Graph Algorithms

Optimal Seam Finding

Dynamic programming algorithm for content-aware image resizing. It finds minimal-energy seams so cropped images keep their visual proportions.

  • Python
  • Dynamic Programming
  • Image Processing

CpG Island Detection in DNA Sequences

Modified max-subarray algorithm that locates CpG nucleotide islands in mammalian DNA, a divide-and-conquer solution to a bioinformatics search problem.

  • Python
  • Divide and Conquer
  • Bioinformatics

Publication

PoPETs 2021 Peer-reviewed

A First Look at Private Communications in Video Games using Visual Features

Read the paper ↗

A. Wajid, N. Kamal, M. Sharjeel, R. M. Sheikh, H. B. Wasim, M. H. Ali, W. Hussain, S. T. Ali, L. Anjum. Proceedings on Privacy Enhancing Technologies, 2021(3), pp. 433–452.

An end-to-end ML study of covert communication channels inside video games — from data labeling and feature engineering through model training and evaluation.

Trajectory

Experience.

Graduate Research Assistant

2025

Georgia Institute of Technology · Shenzhen, China

Urban mobility optimization with predictive ML — +15% prediction accuracy, and SUMO simulation models 30% faster without losing fidelity.

Machine Learning Engineer

2024

Adept Technology Solutions · Islamabad, Pakistan

Built AI for ASCEND, a synthetic-data platform (+25% data quality). Led a full refactor with CI/CD on AWS, cutting deploys from hours to under 15 minutes.

Research Assistant

2020–24

Robotics & Machine Intelligence Lab (ROMI), NUST · Islamabad, Pakistan

Co-authored a PoPETs 2021 paper, and built a full computer-vision stack for surveillance coverage optimization (+20% detection).

Research Intern

2020–21

TUKL-NCAI Deep Learning Lab · Islamabad, Pakistan

Generative and predictive models — Transformers, LSTMs, GANs, Deep Q-Networks — applied to residential load forecasting.

Toolkit

Skills.

Languages

  • Python
  • Java
  • C
  • C++
  • Bash
  • MATLAB
  • SQL

ML / DL

  • PyTorch
  • TensorFlow
  • Scikit-Learn
  • Stable-Baselines3
  • Optuna
  • Hugging Face

Reinforcement Learning

  • Gymnasium
  • OpenSpiel
  • Deep Q-Networks
  • Stable Baselines

CV & Robotics

  • OpenCV
  • ROS / ROS2
  • Gazebo
  • SLAM
  • Kalman / Particle Filters
  • OAK-D

GenAI & NLP

  • LLMs
  • RAG
  • Graph RAG
  • LangChain
  • Graphiti
  • Sentence-BERT
  • spaCy
  • MCP

Data & Databases

  • NumPy
  • Pandas
  • PySpark
  • Neo4j
  • PostgreSQL
  • ChromaDB
  • FAISS

Cloud & DevOps

  • AWS
  • Docker
  • Kubernetes
  • CI/CD
  • FastAPI
  • Pydantic

Hardware & Formal

  • Jetson Nano
  • FPGAs (Vivado)
  • Storm
  • perf
  • Valgrind

Academics

Education.

2024–2026

M.S. Computer Science — Machine Learning

Georgia Institute of Technology

Atlanta, GA

GPA 4.0 / 4.0 · Specialization in Machine Learning

2018–2022

B.E. Electrical Engineering

National University of Sciences & Technology (NUST)

Islamabad, Pakistan

Elected AI specialization stream

Get in touch

Let's build
something.