Final SPI in B.E. Information Technology
AI/ML Engineer · Data Scientist · LLM Apps
Designing AI systems that move from model logic to real product value.
I’m Maitrayee Purohit, an Ahmedabad-based AI/ML engineer and data scientist building practical machine learning products across RAG, NLP, recommendation systems, deep learning, analytics, and full-stack intelligent applications.
- Profile
- AI and data product builder
- Primary stack
- Python, Django, Streamlit
- Availability
- Internships, roles, projects
Portfolio Snapshot
- RoleAI/ML Engineer & Data Scientist
- DomainLLM, NLP, RAG, Deep Learning
- Product layerDjango, Streamlit, analytics UI
- MindsetExplainable, measurable, deployable
Live Profile Dashboard
AI engineering signal at a glance.
RAG, LMS, RL, and algorithm implementation projects
Trustworthiness boost through verification logic
Concurrent students guided across coding, ML, and math
Status
Ready to contribute
AI/ML Engineer
Open
Data scientist roles
Open
Full-time roles
Open
Collaboration
Available
Core Stack
Build-ready toolkit
Python
TensorFlow
Scikit-learn
Django
Streamlit
ChromaDB
FastAPI
OpenAI API
Hugging Face
Pytest
Azure AI
Azure ML
Power BI
GitHub
About
Practical AI engineering and data science with product sense.
AI/ML engineer and data scientist specializing in LLM-powered systems, RAG pipelines, statistical analysis, and agentic workflows. Experienced in designing end-to-end intelligent systems that combine semantic search, vector databases, analytics, and automated verification, with a strong focus on scalable architecture, reliability, and real-world deployment.
My work focuses on clarity: clean data movement, measurable model behavior, understandable outputs, and interfaces that help people trust and use the result.
6+
Core ML algorithms implemented from scratch
40%
Trustworthiness boost in AI research workflow verification
2025
Microsoft Data & AI Skills Internship completed
Skills
A focused stack for building, testing, and presenting AI systems.
Modeling
Supervised and unsupervised learning, deep learning, DQN reinforcement learning, feature engineering, anomaly detection, recommendation systems, statistical analysis, and model evaluation.
Engineering
Python, SQL, TensorFlow, Scikit-learn, NumPy, Pandas, Matplotlib, Seaborn, Pytest, Git, GitHub, environment management, and testing workflows.
Product Apps
Django, Streamlit, FastAPI, Bootstrap, HTML, CSS, SQLite, role-based workflows, analytics dashboards, and interactive demos.
AI Systems
RAG pipelines, agentic workflows, prompt engineering, semantic chunking, vector embeddings, OpenAI API, ChromaDB, SentenceTransformers, Hugging Face, Azure AI Language, Azure ML, NER, text analytics, sentiment analysis, Power BI, data cleaning, and EDA.
What I Build
From prototype to usable AI workflow.
Retrieval-Augmented AI Tools
Document ingestion, embeddings, vector search, citations, verification layers, and useful report output.
Learning & Recommendation Products
Course matching, learner profiles, quiz flows, engagement insights, and explainable recommendation scoring.
ML Foundations & Experiments
Core algorithms, evaluation metrics, reinforcement learning agents, notebooks, and reproducible experiments.
Selected Work
Projects shaped around outcomes, not only models.
01
AI Research Agent
RAG · Verification · StreamlitBuilt a production-style research pipeline that connects web search, document parsing, semantic chunking, embeddings, ChromaDB storage, RAG retrieval, and report generation in one workflow.
40%trustworthiness boost
End-to-endresearch workflow
Testeddependency-aware setup
- Added an automated verification layer to flag unsupported claims and improve trust.
- Created an interactive Streamlit interface with citation-aware output.
- Structured the project with dependency pinning, environment management, and tests.
Python
Streamlit
ChromaDB
SentenceTransformers
RAG
Pytest
02
Smart E-Learning AI Platform
Django · LMS · RecommendationsBuilt an AI-enhanced Learning Management System in Django that supports students, instructors, and admins across course discovery, enrollments, quizzes, certificates, analytics, and personalized learning flows.
3 rolesstudent, instructor, admin
AI-firstquiz, chatbot, recommendations
Productcertificates, analytics, workflows
Added recommendation scoring, learner profiling, quiz generation, chatbot guidance, video-summary support, plagiarism checks, and engagement analysis to show how AI features can be embedded into a real education product.
- Designed role-based workflows for students, instructors, and admins inside one Django application.
- Implemented explainable recommendation and learner-profile logic using progress, quiz, wishlist, and enrollment signals.
- Showcased practical AI integration through chatbot assistance, smart quiz generation, plagiarism checks, and engagement analysis.
Python
Django
SQLite
Recommendation System
NLP
Bootstrap
03
Deep Q-Learning Lunar Lander
RL · TensorFlow · OpenAI GymImplemented a DQN agent for OpenAI Gym’s Lunar Lander using experience replay, target networks, epsilon-greedy exploration, and Bellman-based Q-value updates.
The project focused on teaching the agent to make stable landing decisions through repeated trial-and-error training, balancing exploration with learned policy improvement across episodes.
DQNexperience replay and target network
RLepsilon-greedy exploration
Demointeractive project deployment
Python
TensorFlow
OpenAI Gym
Reinforcement Learning
04
ML Algorithms from Scratch
NumPy · Metrics · FundamentalsDeveloped Linear Regression, Logistic Regression, Decision Trees, K-Means, Anomaly Detection, and Collaborative Filtering from first principles, including strong benchmark outcomes on evaluation datasets.
6+algorithms implemented
Scratchfirst-principles understanding
Benchmarkedevaluated against datasets
Python
NumPy
Scikit-learn
Model Evaluation
Experience & Education
Learning deeply, explaining clearly, applying quickly.
Freelance AI/ML Engineer
- Architected and deployed a production-grade AI Research Agent featuring a full RAG pipeline, from web search and document parsing to semantic chunking, ChromaDB vector storage, and LLM report generation, with an automated claim-verification layer that improved output trustworthiness by 40%.
- Delivered a full-stack Smart E-Learning Platform in Django supporting student, instructor, and admin workflows, with AI-powered course recommendations, NLP-based chatbot assistance, smart quiz generation, plagiarism detection, and learner engagement analytics.
- Trained a Deep Q-Network reinforcement learning agent on OpenAI Gym's Lunar Lander environment using TensorFlow, implementing experience replay, target networks, and epsilon-greedy exploration, achieving stable landing policies and deploying it as a live interactive web demo.
- Re-implemented 6+ core ML algorithms from first principles, including Linear Regression, Logistic Regression, Decision Trees, K-Means, Anomaly Detection, and Collaborative Filtering, then benchmarked them against standard datasets to validate correctness and deepen mathematical understanding.
- Mentored 5+ concurrent students in Python, ML, and mathematics through structured, adaptive learning plans, consistently improving grades from ~70% to 80-90% and simplifying complex topics like gradient descent and statistical inference into intuitive explanations.
Microsoft Data & AI Skills Internship
- Completed Azure AI Fundamentals training and post-training assessment.
- Built NLP workflows using Azure AI Language services for text analytics and sentiment analysis.
- Worked through generative AI workflows in Azure Machine Learning.
- Completed Power BI analysis training with practical visualization exercises.
B.E. Information Technology
- Final SPI: 10 / 10.
- Focused on AI/ML systems, full-stack problem solving, and hands-on implementation.
- Developed projects across ML, RL, analytics, and applied LLM workflows.
Machine Learning, Python & Computer Vision
- Completed Machine Learning Specialization by DeepLearning.AI and Stanford University, taught by Andrew Ng.
- Earned Python certification from GUVI with IIT certification recognition.
- Built a Face Recognition Application using Python as part of GUVI's AI-for-India event.
Contact
Looking for an AI/ML engineer or data scientist who can build and communicate?
I’m open to internships, full-time roles, data scientist opportunities, and collaborations where machine learning or analytics needs to become a clear, useful product experience.