About Me

I'm Abdelazim Lokma, I recently graduated from Boston University with a Master's degree in A.I. with a GPA of 3.8/4.0. My background includes a Bachelor's in Computer Science with a minor in Business Administration. I have a passion for AI-driven solutions, including machine learning models and computer vision projects!

Nationality: Saudi Arabia

Abdelazim Lokma

Education

Boston University, Boston, MA

Master of Science in Artificial Intelligence | GPA: 3.8/4.0 | Sep 2023 - Jan 2025

Key Coursework:

  • Spark! ML X-Lab Practicum (A)
  • Advanced Topics in Computer Science (A)
  • Graduate Databases (A)
  • Machine Learning (A-)
  • Computer Vision (A-)
  • Natural Language Processing (A)

Boston University, Boston, MA

Bachelor of Arts and Sciences in Computer Science | May 2023

Minor in Business Administration

Experience

A.I Engineer Intern

Horizon World Salon Inc., Remote (Based in New York, NY) | March 2025 - Present

  • Developed a Retrieval-Augmented Generation (R.A.G) pipeline to recommend experts for specific events, enhancing matchmaking accuracy.
  • Collaborated with cross-functional teams to integrate A.I solutions, improving event curation and attendee satisfaction.
  • Contributed to data preprocessing and feature engineering, enhancing the performance of machine learning models in expert recommendation.

A.I Engineer

Talentora, Boston, MA | Jan 2025 - Present

  • Contributed to the creation of Talentora, a BU-based startup focused on replacing traditional resume screening with scalable, AI-powered interviews.
  • Designed and improved Talentora's AI-powered interview system, using LangChain and Pipecat Flows for smooth, automated interviews.
  • Created an ensemble SVR model that processes outputs from three Hume models (Prosody, Language, and Face) through separate SVR models, with a final regression model to weight and combine predictions.
  • Implemented advanced model optimization techniques to prevent data leakage and ensure robust performance in the meta regression model.

Teaching Assistant - CS 611 Graduate O.O.P & Software Design in Java

Boston University, Sep 2024 - Dec 2024

  • Taught and mentored over 50 graduate students in lab sessions, providing guidance on Object-Oriented Programming concepts in Java.
  • Collaborated with the course instructor to influence course design, helping to shape lab exercises and assignments to better align with learning objectives.

Projects

  • FairCase A.I Assistant

    Designed FairCase, an AI-powered tool for the National Lawyers Guild Massachusetts Chapter, automating the analysis of civilian complaints against Boston Police Department officers. Implemented a custom LangChain multi-query retriever that dynamically generates RAG queries for each allegation-description pair.

  • Sign Language Identification Program

    Spearheaded design and implementation of an algorithm to recognize American Sign Language (ASL) gestures using template matching and circularity measurement techniques. Achieved a greater than 90% classification accuracy in ASL gesture recognition.

  • Fully Convolutional Network for Semantic Segmentation

    Implemented and trained a Fully Convolutional Network (FCN-8s) with a VGG-16 backbone for semantic segmentation, optimizing key components like the forward pass and loss function. Achieved mIoU > 27% on a custom dataset by fine-tuning the model's hyper-parameters.

  • ARC AGI Kaggle Competition

    Designed and implemented an ensemble solution for the ARC Challenge, combining reinforcement learning-based LLMs with fine-tuned Instruct LLaMA models, achieving competitive performance in abstract reasoning tasks.

  • Pacathon 2024 RL Hackathon

    Designed and implemented advanced heuristics for Pacman AI, optimizing neural network behavior and achieving high performance in the competition.

  • Financial (AI)d

    Attended CalHacks 11.0! Designed a speech enabled AI Chat Bot that provides financial assistance & education on stock market investing.

  • Weather Prediction Model

    Developed an RNN & Ridge Regression based models using historical weather data to predict future temperatures across multiple cities, achieving a mean absolute error of 2°F from ground truth results.

  • Multi-Vehicle Tracking System

    Implemented a Kalman filter for single-object tracking and a data association algorithm to assign unique IDs to multiple objects, enabling robust object tracking in video sequences.

  • Simple DBMS

    Developed a custom DBMS in C++ with core components like heap files for storage, buffer pool for caching, and optimized data retrieval using B+Tree indexing.

  • Predicting Amazon Reviews

    Implemented a Gradient Boosting Classifier to predict Amazon review ratings using sentiment analysis on a dataset of 130,000 reviews.

  • Fit-Gen

    Co-created an Android app to catalog users' clothing inventory and generate personalized outfits based on fashion principles and external factors.

  • Smart Pantry

    Designed a Java program to track expiry dates of groceries and notify users when items are soon to expire. Added a recipe book that recommends recipes using items near expiration.

Skills

  • Programming Languages

    Java, Python (Pandas, PyTorch, HuggingFace, NumPy), LangChain, C++, SQL, Git

  • Cloud & DevOps

    AWS Cloud Practitioner (CLF-02) Certified, Docker

  • Languages

    Arabic (Native), English (Fluent), French (Conversational)

Resume

Contact Me

Feel free to get in touch with me for collaborations or opportunities!

Email: lokma.abdelazim@gmail.com

Phone: U.S.A: +1 (678) 818-2286 | Egypt: +20 (100) 608-3335

LinkedIn: www.linkedin.com/in/abdelazim-lokma