Melisa Sever
AI Engineer & Backend Engineer
Architecting high-performance AI ecosystems and backend infrastructures that turn complex models into seamless, production-ready user experiences.
Download ResumeAbout Me
Hi, I'm Melisa. I'm an AI & Backend Engineer with 3 years of experience building scalable ML systems and APIs on Google Cloud. I specialize in bridging the gap between experimental AI and production reality. Whether I'm architecting LLM-powered platforms or optimizing inference pipelines to slash latency by 55%, my focus is on turning deep learning and NLP concepts into high-performing, reliable applications. I love tackling architectural bottlenecks and building intelligent systems that deliver immediate, real-world value.
Experience
AI Engineer
Reality AI Labs, San Francisco, California
- Architected and deployed LLM-powered syllabus generation platform using LangChain, Gemini, Vertex AI, serving real educator workflows.
- Reduced generation latency by 55% and improved output coherence by 40% by redesigning prompt pipelines and the inference flow.
- Built parallelized image prompt generation, automatic theme detection, increasing content production speed and visual consistency.
Backend Engineer
Vibing Ventures, San Francisco, California
- Built and deployed an AI-powered quiz-generation platform used by 200+ students, generating personalized assessments and feedback.
- Developed scalable backend services using Python, FastAPI, LangChain, and Vertex AI.
- Implemented automated evaluation pipelines CI/CD on Google Cloud, maintaining 99% uptime.
Frontend Developer
Pixagor Data Center and Software Services Inc., Istanbul, Türkiye
- Developed a healthcare appointment web app, increasing online bookings 50%.
- Improved user satisfaction 60% through performance optimization and async API redesign.
- Delivered production features within Agile/Scrum teams.
Education
Bachelor of Science in Computer Science
San Francisco State University, San Francisco, CA
Featured Projects
Secret & API Key Detector
A lightweight, zero-dependency Python tool that scans your codebase for hardcoded secrets, API keys, tokens, and passwords before they end up on GitHub.
View Project →Log Analyzer
A Python tool that parses server log files and detects suspicious behavior, including brute-force attacks, SQL injection, XSS attempts, bot activity, IP anomalies, and repeated access patterns. Comes with a CLI, email/Slack alerting, and an interactive browser dashboard.
View Project →Satellite Onboard AI Simulation
A lightweight Python project simulating edge computing for Earth observation satellites, inspired by ESA's φ-sat-2 mission. This system demonstrates how artificial intelligence can be deployed onboard satellites to autonomously filter image data, reducing bandwidth consumption and extending mission capabilities.
View Project →Space Weather Forecasting & Anomaly Detection System
AI-powered space weather monitoring system for protecting satellites and astronauts from solar radiation
View Project →LangGraph Interactive Fiction Engine
A powerful, stateful interactive fiction (choose-your-own-adventure) engine built with LangGraph and Llama 3.1 (via Groq). This system generates dynamic, branching narratives that adapt to user choices while maintaining deep story consistency and coherence.
View Project →Satellite Image Analysis Project
I've created a complete AI toolkit for satellite image analysis. It's a fast, production ready system that can detect wildfires by identifying active fire zones and mapping burn scars, monitor floods by tracking water extent and assessing their impact, and measure urban sprawl by analyzing how cities grow over time.
View Project →Space Weather Forecasting & Anomaly Detection System
AI-powered space weather monitoring system for protecting satellites and astronauts from solar radiation
View Project →Advanced Bot Detection System
A production-ready, AI-powered bot detection system with deep learning, SQLite persistence, CAPTCHA integration, and FastAPI.
View Project →LLM API Abuse Detection & Shield
A production-ready abuse detection system for LLM and ML APIs that detects prompt attacks, scraping, and data exfiltration attempts in real-time.
View Project →AI-Powered Threat Intelligence & SOC Assistant
A comprehensive cybersecurity SaaS tool that ingests logs, detects anomalies using machine learning, and generates human-readable incident reports with AI-powered recommendations.
View Project →Medical Jargon Translator
A Python application that translates confusing medical terminology from doctor's notes and lab reports into plain, easy-to-understand English using AI. Ask follow-up questions, save your translation history, and export reports for easy sharing.
View Project →Heart Attack Prediction-Machine Learning Pipeline
A comprehensive end-to-end machine learning project for predicting heart attack risk using patient medical data. This project demonstrates a complete ML workflow from data exploration to model evaluation.
View Project →CDC Diabetes Classification Project
A comprehensive machine learning project demonstrating how data preprocessing, feature scaling, and hyperparameter tuning improve predictive performance for diabetes classification.
View Project →Medical Cost Prediction System
An interactive machine learning web application that predicts healthcare costs based on patient demographics and health factors. Built with Python, Scikit-learn, and Streamlit.
View Project →AI-Powered Stock Prediction Dashboard
A lightweight, free-to-use stock market analysis and prediction tool built with Python. Features machine learning price predictions, sentiment analysis from news, and technical indicators visualization, all without requiring paid API keys!
View Project →Personalized Shopping Recommender System
A hybrid recommendation system that combines collaborative filtering and sentiment analysis to provide personalized product recommendations. Built with Python (scikit-learn) and features an interactive web dashboard.
View Project →AI Shopping Research Agent
A free AI-powered shopping assistant that helps you find the best products based on your requirements. Uses Groq's free API for lightning-fast AI responses and requires no credit card.
View Project →Carbon Footprint Tracker App
A comprehensive web application for tracking, analyzing, and reducing your carbon footprint using Machine Learning and AI-powered suggestions.
View Project →Customer Support Ticketing System Simulator
A multi-agent system built with LangGraph that automatically classifies, routes, and resolves customer support tickets using specialized AI agents.
View Project →Multi-Destination Travel Itinerary Planner
An advanced AI travel planning system with coordinated specialized agents that generate comprehensive, conflict-free itineraries for multi-destination trips.
View Project →Code Review & Bug-Fixing Swarm
A multi-agent AI system for automated code analysis, security auditing, and quality improvement across 8 programming languages.
View Project →Autonomous Deep Research Agent
A sophisticated multi-agent AI system that conducts comprehensive research on any topic using autonomous agents, self-reflection, and human-in-the-loop oversight.
View Project →Credit Card Fraud Detection
A lightweight, zero-dependency machine learning system designed to detect fraudulent credit card transactions.This project uses Logistic Regression and Stochastic Gradient Descent, written in native Java, to classify transactions as Safe or High Risk.
View Project →Autonomous 3D Robot Fleet Simulation
A simulation environment for studying the behavior and coordination of an autonomous 3D robot fleet, focusing on pathfinding and task allocation algorithms.
View Project →ShadowVault-Zero-Trust API Honeypot
A Python honeypot that mimics banking APIs to deceive attackers. Detects 5+ attack types, profiles sessions, and generates automated IOCs every 5 minutes with real-time analytics.
View Project →AI-Enhanced Threat Detection System
A real-time cybersecurity threat detection system using machine learning to identify and automatically respond to authentication attacks, application-level attacks, and insider threats.
View Project →FraudGuard Pro-Real-Time Bank Fraud Detector
An intelligent fraud detection system that monitors transactions in real time, automatically flags suspicious activity, and sends instant SMS alerts built with pure Python and SQLite.
View Project →Swarm Robotics Simulation
A simulation environment for studying the collective behavior of swarm robots. Explores decentralized coordination algorithms for tasks like foraging and area coverage.
View Project →Autonomous Vehicle Simulator
A lightweight 2D self-driving car simulation built with Python and Pygame. Features realistic path planning, obstacle avoidance, traffic light detection, and LiDAR sensors.
View Project →AI-Powered Resume Screener
A Python resume-matching system that parses resumes and job postings using semantic similarity. Applies multi-factor scoring via TF-IDF and configurable weighting, integrates SQLite for batch processing.
View Project →F1 Predictions 2025
A machine learning project that uses historical Formula 1 race data and the FastF1 API to predict driver and team outcomes for the 2025 F1 season.
View Project →Medical Note AI Assistant
An intelligent system that transcribes voice recordings into structured SOAP-format clinical notes, supporting 6 languages with persistent database storage.
View Project →School Nurse Triage System
A rule-based AI triage assistant that helps school nurses rapidly classify student health concerns into urgent vs. non-urgent using official pediatric protocols.
View Project →Text-to-Music Generator
An offline AI music generation tool that creates custom royalty-free tracks from text prompts ideal for content creators, podcasters, and video editors.
View Project →AI Video Generator
A Python-based tool for generating professional videos complete with background music, AI voice-overs, and stunning visual effects from simple inputs.
View Project →AI-Driven Threat Detection System
This project is an AI-driven threat detection system designed to analyze network traffic and identify potential anomalies. It uses a machine learning model (Isolation Forest) to detect suspicious activities within a network by distinguishing between normal and abnormal traffic patterns.
View Project →AI-Powered Cybersecurity Threat Detection System
This project implements an AI-powered system to detect cybersecurity threats across three domains: network anomalies, malware, and phishing emails. It uses machine learning models from scikit-learn, including Isolation Forest, Random Forest, and an MLP-based autoencoder, to identify potential threats in synthetic data.
View Project →Skills
Languages
AI/ML & NLP
Frameworks & Libraries
Cloud & DevOps
Soft Skills
Contact Me
I'm always open to new opportunities and collaborations. Feel free to reach out!
melisasever2000@gmail.com (650) 771-8750