RAG Document Copilot
A FastAPI service that answers questions about PDFs with Cohere embeddings and ChromaDB search. It returns grounded answers with page citations, refuses unsupported questions, and includes mocked pytest coverage.
Mohammad Hashemi
Applied AI & AI Software Engineer
I build grounded LLM applications, machine-learning systems, and AI-enabled robotics software—backed by backend engineering, research, and product development.
My recent work includes a citation-aware RAG service for PDF question answering, a tool-calling FPL assistant using live API data, and an AI-assisted mixed-reality interface for programming a UR10 robot. I bring experience across Python, FastAPI, machine learning, robotics, XR, and released Android software.
Selected work
Applied AI systems
Three projects showing grounded retrieval, live-data tool use, and structured AI interaction with a physical robot.
FPL Agent
A tool-calling assistant that plans multi-step answers to Fantasy Premier League questions using live FPL API data, player and fixture tools, and inspectable tool-call traces.
AURaPath
A mixed-reality UR10 trajectory-authoring system with digital-twin preview and a voice/LLM pipeline that turns free-form instructions into structured robot commands behind deterministic keyword-based safety checks.
Capabilities with evidence
Models connected to reliable software
My work focuses on the engineering around model behavior: evidence, APIs, tools, structured interfaces, evaluation, and user workflows.
Grounded LLM applications
Retrieval, embeddings, vector search, citations, refusal behavior, and inspectable tool execution.
See RAG Document Copilot and FPL AgentAI-enabled robotics
Natural-language interaction, structured robot commands, mixed-reality authoring, digital twins, and deterministic command checks.
See AURaPathMachine learning
Customer segmentation and purchase-prediction pipelines using SVM, Random Forest, and XGBoost.
See the Audiobook ML projectProduct and backend engineering
FastAPI services, REST integrations, testing, Android architecture, and delivery of a released software product.
See OfficeMeter and API-based projectsSelected research
Human-robot interaction for UR10 programming
AURaPath connects mixed-reality trajectory authoring, digital-twin preview, natural-language commands, and deterministic checks before robot execution. Related work was published at IEEE CCECE 2026.
- Area
- AI, robotics & HRI
- System
- HoloLens 2 + UR10
- Venue
- IEEE CCECE 2026
Supporting engineering
Software delivery and problem solving
Released product work, backend development, research, robotics, and algorithms strengthen how I build and evaluate AI systems.
Released Android application
OfficeMeter
A product for tracking and planning office attendance, demonstrating user-facing software architecture and release delivery.
Research & algorithms
Computational geometry and competitive programming
Peer-reviewed research, an MSc thesis in computational geometry, ICPC regional results, and robotics competitions provide a rigorous problem-solving foundation.
Get in touch
Building an applied AI or robotics team?
Explore the project evidence, review my code, or contact me to discuss engineering opportunities.