AI Trainee Engineer Intern @ Analytiverse
BS Computer Science Graduate · University of Education, Lahore · Class of 2026
About · Engineering Focus · Projects · Tech Stack · Contact
I'm Hammad Ali Tahir, a Computer Science graduate from the University of Education, Lahore, currently working as an AI Trainee Engineer Intern at Analytiverse.
I build AI applications across Machine Learning, Natural Language Processing, Retrieval-Augmented Generation, and agentic workflows. My projects connect the full application: preparing documents, retrieving relevant evidence, coordinating model and tool calls, serving streaming responses, and building the interface around them.
My main engineering interest is making AI answers more dependable. That means improving retrieval, checking whether the evidence supports a response, handling unsuccessful searches, and tracing what happened when a workflow goes wrong.
Current focus: Generative AI applications, retrieval quality, tool-using agents, and Python backend engineering.
| Focus | How I apply it |
|---|---|
| Agentic Workflows | Build LangGraph workflows with query routing, task decomposition, tool calls, explicit state, and bounded retrieval retries. |
| Retrieval-Augmented Generation | Combine dense and keyword search with cross-encoder reranking, chunk grading, query rewriting, and source citations. |
| Full-Stack AI Applications | Connect FastAPI backends and Server-Sent Events to React interfaces, authentication, and stored conversations. |
| Evaluation & Observability | Evaluate retrieval and answer quality with RAGAS, and inspect execution traces with LangSmith. |
| Applied Machine Learning | Build preprocessing pipelines, engineer features, compare classifiers, and evaluate predictive performance. |
| Project | Engineering & Architecture | Capabilities | Explore |
|---|---|---|---|
| UOE AI Assistant Academic RAG application · Final-year project |
Hybrid retrieval with dense search and BM25, cross-encoder reranking, query rewriting, chunk grading, and retry logic. FastAPI + SSE backend connected to React, Pinecone, OpenAI, and Supabase. | Answers questions about university curricula and regulations with source-grounded retrieval. Includes conversation history, RAGAS evaluation, and LangSmith tracing. | Open Assistant |
| AI Legal Case Manager Uraan AI Techathon 1.0 |
Ensemble machine learning for case classification and priority prediction, combined with embedding-based retrieval of relevant judgments. Built with Python, scikit-learn, and Streamlit. | Combines case categorization, triage, and precedent search in one application. | Open Demo |
| GitHub Research Agent Tool-assisted repository research |
LangGraph ReAct workflow with GitHub tool integrations, repository inspection, caching, and a React + Vite chat interface. | Retrieves repository metadata, examines issues and code, and synthesizes findings into research responses. | Browse Projects |
| Loan Approval Predictor End-to-end machine learning |
Compares six classifiers with column-transformer preprocessing, feature engineering, custom logging, and health monitoring. Packaged with Docker for deployment on Render. | Connects data preparation, model selection, and prediction serving in a complete application. | View Repository |
Inside UOE AI Assistant — retrieval, reasoning, and evaluation
The workflow routes requests, decomposes questions when necessary, and selects the retrieval path needed to find supporting information.
Dense search and BM25 keyword search retrieve candidate passages. A cross-encoder reranks them, and chunk grading checks their relevance. Query rewriting and bounded retries help when the initial search does not find enough useful context.
The FastAPI backend streams responses through Server-Sent Events to the React frontend. Supabase supports authentication and conversation history, while source references help users inspect the basis of an answer.
RAGAS supports retrieval and answer evaluation. LangSmith traces make it possible to inspect intermediate steps and investigate failed or slow requests.
Cricket fast bowler and former department team captain. I enjoy historical and geopolitical documentaries, and bring the same interest in strategy and problem-solving to my engineering work.
| Find me | Details |
|---|---|
| Current Role | AI Trainee Engineer Intern at Analytiverse |
| Education | BS Computer Science · University of Education, Lahore · 2026 |
| Interactive Portfolio | fedora-portfolio.vercel.app |
| Hammad Ali Tahir | |
| hammadalitahir8@gmail.com | |
| GitHub | @HammadAli08 |
AI = Logic + Data + Imagination.