Meer Musabih Saleem

Professional Summary

AI/ML-focused Software Engineering student with hands-on machine learning internship experience building and evaluating end-to-end ML systems: a Random Forest ranking model (Precision@20 0.70 vs 0.40 baseline), an NLP toxicity classifier (~86.4% accuracy, 0.950 ROC-AUC), LLM-powered AI agents, voice AI, and demand forecasting. Applied Python ML stack (scikit-learn, Pandas, NumPy, FastAPI) across predictive analytics, NLP, ranking models, feature engineering, generative AI, LLM applications, and conversational AI.

Technical Skills

AI / Machine Learning: Machine Learning, Predictive Analytics, NLP (Natural Language Processing), Classification, Ranking Models, Feature Engineering, Model Evaluation, Generative AI, LLM Applications, AI Agents, Conversational AI, Voice AI

Python / ML Stack: Python, Pandas, NumPy, scikit-learn, FastAPI

LLMs / AI Tools: Ollama, Qwen, Llama, Anthropic / Claude, Microsoft Copilot Studio, n8n

Development / Data: Git, GitHub, REST APIs, JSON, Data Analysis

Software Engineering: C++, JavaScript, React, Node.js, Express, MongoDB (MERN), Flutter / Dart

Experience

Machine Learning Intern — FlyRank AI Content-Refresh Ranking Capstone

Built a Random Forest ranking model (scikit-learn, 300 trees, class-balanced, numeric + categorical features) over ~30,000 content-page records from the internship data warehouse to rank pages by priority for human content-refresh review.
Achieved Precision@20 of 0.70 and Precision@50 of 0.68, versus a 0.40 / 0.36 rule-based baseline, using a client-grouped train/test split that prevented client leakage between training and testing.
Engineered a 32-feature modeling set from 44 original columns, excluded target and leakage variables, and surfaced top predictive signals — 90-day impressions, average search position, days with impressions, content age, character/word count, CTR, pageviews, scroll rate.
Delivered a ranked review queue with priority tiers and reason codes as decision support, helping human reviewers identify pages likely to deserve a content refresh.

AI / ML Projects

PersonaGuard AI — AI Recruitment Intelligence Python · NLP · Sentence Transformers · Logistic Regression · FastAPI

Built an AI recruitment-intelligence system that analyzes publicly available social-media information to assist recruitment screening using NLP and machine-learning classification.
Trained a Logistic Regression toxicity classifier on Sentence Transformer embeddings, reaching ~86.4% accuracy and 0.950 ROC-AUC; served inference via a FastAPI/Uvicorn API with rigorous model evaluation.

AI Job Finder Agent n8n · AI Agents · REST APIs · Ollama · Qwen 2.5 1.5B · JSON

Architected an AI-agent pipeline that turns natural-language job preferences into ranked matches: searches live public job APIs, retrieves and deduplicates listings, and evaluates each posting against user criteria with a locally hosted LLM (Qwen 2.5 1.5B via Ollama).
Returns structured JSON rankings of suitable opportunities; processed 275 listings → 271 after deduplication → 6 final matches, built entirely on free/local tooling with no paid API keys.

AI Call Center — Voice AI / Conversational AI Python · Twilio · Speech-to-Text · LLM · Text-to-Speech

Built an AI-powered call-center system integrating speech-to-text, LLM-driven conversational intelligence, and text-to-speech over Twilio telephony for real-time voice interaction, automated customer interaction, and call routing.

Inventory Forecasting & Predictive Analysis Python · Pandas · NumPy · scikit-learn

Developed an applied ML / predictive-analytics system for multi-outlet retail businesses, forecasting product demand from current and historical inventory/sales data and identifying fast-moving products and sell-out risk.
Applied feature engineering and supervised learning to generate outlet-specific recommendations on which products to stock or drop per outlet, supporting inventory planning and demand forecasting.

Microsoft Copilot Studio — Enterprise AI (Ongoing) Conversational AI · Workflow Automation

Currently building AI copilots for enterprise scenarios — conversational AI assistants and AI-assisted business-process/workflow automation within Microsoft Copilot Studio.

Additional Projects MERN · C++ · Flutter

SoleMate — full-stack MERN e-commerce shoe store (React, Node.js, Express, MongoDB)
C++ Space Shooter — single-file arcade game built in C++
Campus Companion — Flutter/Dart mobile campus companion application

Professional Summary

Technical Skills

Experience

AI / ML Projects

Education

Certifications / AI Training