Hi, I'm

MarijaManeva

DataScience·AIEngineer
LLM Applications
Scroll
About Me

Who I Am

Marija Maneva

Marija Maneva

Data Science · AI Engineer

Download my CV

Bio

As an AI Engineer with 2 years of experience and a background in Data Science, I specialize in automating complex tasks and systems using AI. I work with Python, LLMs, LangChain, LlamaIndex, vector databases, and RAG architectures to create impactful solutions. Detail-oriented and highly motivated, I'm passionate about AI's transformative potential and committed to building solutions that deliver real value to clients and end-users.

Education

2023 – 2025

MSc. Computer Engineering (Data Science)

University of Pavia · Italy

Thesis: "A Deep Learning Framework for Vector Graphics Generation using Stable Diffusion"

2019 – 2022

BSc. Computer & Electrical Engineering

University of Pavia · Italy

Thesis: "Machine Learning Techniques for the Prediction of the Progression of Multiple Sclerosis"

Experience

June 2025 – Present

AI Engineer – Consultant

Reply · Fulltime

Dec 2023 – March 2025

Data Science – AI Engineer

Produkto · Part-time

Languages

ItalianNative
EnglishC1
GermanA1

Skills & Technologies

Core Skills

RAGMulti-Agent SystemsPrompt EngineeringFew-Shot LearningProblem SolvingSystem DesignData AnalysisPublic SpeakingPresentation DesignWorkshop Creation

AI & LLM

Claude CodeLangChainLangGraphLlamaIndexLiteLLMLangFuseRAGASCodexGitHub CopilotGeminiChatGPTClaudeNotebookLMGemini CLI

Cloud & Infrastructure

AWSGCPBigQueryCloud RunPostgreSQLNeo4JMongoDBVercelGradioStreamlit

Programming & Markup Languages

PythonSQLHTML/CSS

Frameworks & Libraries

Google ADKGemini SDKVertex AI SearchTensorFlowPyTorchScikit-learnKerasXGBoostFastAPIFlaskPySparkHadoopMongoDBPandasNumPyOpenCVMatplotlib

Personal Projects

langlearn-v5b5.onrender.com
LangLearn — The Adaptive Language Course preview
Live

LangLearn — The Adaptive Language Course

Adaptive language course that re-reads your level after every answer, weaving theory, exercises, spaced review, vocabulary and CEFR tests into one path that never wastes your time.

Next.jsReactTypeScriptSupabase
support-survival.onrender.com
StatMed — Mortality Risk & Survival Analysis preview
Live

StatMed — Mortality Risk & Survival Analysis

Clinical prediction system for seriously ill hospitalized patients (SUPPORT cohort). Combines classical survival analysis (Kaplan-Meier, Cox PH) with ML baselines, focusing on calibration, fairness, and reproducibility. Served via FastAPI with an interactive risk predictor and a risk-stratified triage panel.

PythonSurvival Analysisscikit-learnFastAPIDockerRender
langlearn-v5b5.onrender.com
LangLearn — The Adaptive Language Course preview
Live

LangLearn — The Adaptive Language Course

Adaptive language course that re-reads your level after every answer, weaving theory, exercises, spaced review, vocabulary and CEFR tests into one path that never wastes your time.

Next.jsReactTypeScriptSupabase
support-survival.onrender.com
StatMed — Mortality Risk & Survival Analysis preview
Live

StatMed — Mortality Risk & Survival Analysis

Clinical prediction system for seriously ill hospitalized patients (SUPPORT cohort). Combines classical survival analysis (Kaplan-Meier, Cox PH) with ML baselines, focusing on calibration, fairness, and reproducibility. Served via FastAPI with an interactive risk predictor and a risk-stratified triage panel.

PythonSurvival Analysisscikit-learnFastAPIDockerRender
langlearn-v5b5.onrender.com
LangLearn — The Adaptive Language Course preview
Live

LangLearn — The Adaptive Language Course

Adaptive language course that re-reads your level after every answer, weaving theory, exercises, spaced review, vocabulary and CEFR tests into one path that never wastes your time.

Next.jsReactTypeScriptSupabase
support-survival.onrender.com
StatMed — Mortality Risk & Survival Analysis preview
Live

StatMed — Mortality Risk & Survival Analysis

Clinical prediction system for seriously ill hospitalized patients (SUPPORT cohort). Combines classical survival analysis (Kaplan-Meier, Cox PH) with ML baselines, focusing on calibration, fairness, and reproducibility. Served via FastAPI with an interactive risk predictor and a risk-stratified triage panel.

PythonSurvival Analysisscikit-learnFastAPIDockerRender
langlearn-v5b5.onrender.com
LangLearn — The Adaptive Language Course preview
Live

LangLearn — The Adaptive Language Course

Adaptive language course that re-reads your level after every answer, weaving theory, exercises, spaced review, vocabulary and CEFR tests into one path that never wastes your time.

Next.jsReactTypeScriptSupabase
support-survival.onrender.com
StatMed — Mortality Risk & Survival Analysis preview
Live

StatMed — Mortality Risk & Survival Analysis

Clinical prediction system for seriously ill hospitalized patients (SUPPORT cohort). Combines classical survival analysis (Kaplan-Meier, Cox PH) with ML baselines, focusing on calibration, fairness, and reproducibility. Served via FastAPI with an interactive risk predictor and a risk-stratified triage panel.

PythonSurvival Analysisscikit-learnFastAPIDockerRender
Completed

Master Thesis — Stable Diffusion for Vector Graphics

Developed a full AI generation system from text prompts with complete data pipeline. Fine-tuned Stable Diffusion on a custom dataset, deployed inference endpoint, and implemented AI-assisted prompt refinement with a user-friendly interface.

Stable DiffusionFine-tuningDeep LearningPython
Completed

Movie Recommendation App

Conversational recommendation application using generative AI for personalized movie suggestions through natural language. Showcases RAG architecture and multi-source data integration.

ChromaDBSentenceTransformerLangChainOpenAITMDB APIGradio
View on GitHub
Completed

Large-Scale Movie Reviews Analytics

Conducted large-scale data analysis on movie reviews using big data technologies. Demonstrated distributed computing and cross-platform processing.

HDFSMapReduceSparkMongoDBPandasPython
View on GitHub
Completed

Cloud Computing — Scalable Distributed App

Developed a distributed, scalable, elastic, and fault-tolerant application leveraging AWS cloud services. Production-grade microservices with serverless architecture and infrastructure automation.

AWSMicroservicesServerlessInfrastructure Automation
View on GitHub

Professional Projects

Projects I worked on professionally
the code isn't public as it contains sensitive company data.

Completed

AI-Powered Logo Generation Tool

Implemented an AI-powered tool for automated logo generation at Reply, including fine-tuning a deep learning model on a logo dataset and integrating a vectorization pipeline to convert raster images into resolution-independent SVG formats.

AI
Generated SVGs
E2E
Pipeline
Prod
Deployed
Deep LearningFine-tuningSVG VectorizationPython
Completed

Skill Agent — Conversational Profiling Tool

Developed a conversational profiling tool for SDLC work at Produkto using cloud-friendly services. Built with PostgreSQL + SQLAlchemy/Alembic for multi-user support, GPT-4 via LiteLLM, delivery flow tracking, and Angular frontend.

Multi
User Support
GPT-4
Powered
Full
Stack
LangChainGPT-4PostgreSQLAngularLiteLLM
Completed

Proposal Validation Agent

Built a full AI pipeline for automated checking of business proposals. Implemented specialized agents for rules validation, tag extraction, conflict detection, and proof-of-work using Python, FastAPI, OpenAI, LangChain, LangFuse, and RAGAS for evaluation.

5
Specialized Agents
Auto
Validation
RAGAS
Evaluated
LangChainFastAPIOpenAIRAGASLangFuse
Completed

Business Analyst Chatbot

Created an intelligent chatbot with a hybrid memory system combining short-term and long-term storage for smooth, context-aware conversations. Designed for business analyst workflows with persistent context across sessions.

Hybrid
Memory
Persist
Context
Multi
Turn Chats
LLMMemory SystemsLangChainPython
Completed

File Processor & Parser Pipeline

Built a document classification and parsing system that handles PDF, DOCX, and PPTX files using PyMuPDF, python-docx, and Pillow. Features meaning-based document sorting with flexible GPT model selection (Vision or Text mode).

3
File Formats
Vision+
Text Modes
Auto
Classification
PyMuPDFGPT VisionPythonpython-docx
Get in Touch

Let's Connect

Open to opportunities and collaborations. Feel free to reach out!

Contact Info

Email

marijamaneva99@gmail.com

GitHub

marijamaneva

Usually within 24 hours