Syed Alam Shah Bukhari
AI Engineer building RAG systems LLM Production Agents
New to Contra
Syed Alam is ready for their next project!
AI Equity Research Tool for the Pakistan Stock Exchange
An AI system that automates equity research for PSX listed companies (starting with ENGRO, HUBC, MCB), ingesting annual reports and generating cited, analysis backed answers instead of requiring investors to read hundreds of pages manually.
Instead of a simple chatbot, this is built as an orchestrated AI pipeline: documents are ingested, chunked, embedded, and retrieved based on meaning rather than just keywords, then grounded into answers with source citations, covering things like risk factors, growth drivers, and financial performance.
What it does:
• Ingests financial reports and structures them for retrieval
• Answers natural language questions with citations back to the source document
• Uses an AI orchestration pipeline (LangGraph) rather than a single linear prompt, so it can be extended with retries and multi step reasoning
• Built for evaluation from day one, not just whether it answers, but whether the answer is accurate and grounded
Stack: Python, LangGraph, Gemini API, ChromaDB, FastAPI (in progress)
Currently in active development. The core retrieval pipeline is working end to end, with automated evaluation and a production API layer underway next.
0
55
AI-Powered Browser Extension for Enterprise ERP Automation
Built a Microsoft Edge extension (Manifest V3) that uses LLM-driven decision-making to automate repetitive workflows inside Oracle EBS ERP starting with Purchase Requisition creation.
The extension reads the page, decides the next action using an LLM, and handles form-filling and navigation automatically turning a multi-step manual process into a guided, AI-assisted flow.
Stack: Node.js, Edge Manifest V3, LLM integration for decision-making
Delivered to a paying client and currently in testing for expanded workflows.
0
51
I build production-oriented Retrieval-Augmented Generation (RAG) systems that connect LLMs with your private documents and knowledge bases.
What I can build:
• Document ingestion and intelligent chunking
• Embeddings and semantic/vector search
• ChromaDB / Pinecone / pgvector integration
• LLM-powered question answering
• Retrieval optimization and relevance improvement
• Source-grounded responses to reduce hallucinations
• FastAPI backend and API integration
• Dockerized, maintainable architecture
Example: Built a Document Q&A system using Gemini + ChromaDB that retrieves relevant document context and generates grounded answers.
Ideal for companies looking to turn internal documents, knowledge bases, PDFs, or technical content into an AI-powered search and Q&A system.
0
33
Feature Lead on the COA pipeline at Attuned AI (PM Accelerator) built hybrid RAG retrieval, an LLM extraction API, and a 100-case evaluation harness. Promoted from intern to feature lead, coordinated 2 engineers.
0
48