Agentic RAG · Legal · Energy · Government
Production RAG for regulated industries.
I build RAG systems for legal, energy, and government teams. The hardest problem isn't the LLM. It's building where compliance and auditability aren't nice-to-haves. They're the architecture.
Selected case studies
Work
Production AI systems shipped for legal, energy, and SaaS teams. Each case study includes the technical approach, the outcome, and why it mattered.
Malakah: Agentic RAG Legal Assistant
AI legal assistant achieving 98% retrieval accuracy on Saudi regulations, contributing to $600K in pre-seed funding.
Enterprise AI Search and Chatbot: Energy Sector
Intelligent search system processing 300+ GB of unstructured enterprise data for a major energy company, including 1K–6K page technical documents.
Multi-Tenant AI SaaS Platform: Document Intelligence
AI-powered document intelligence platform with multi-tenant architecture, RBAC, and hybrid retrieval.
Bilingual Voice Agent on LiveKit
Production-grade real-time voice agent supporting English and Urdu, with sub-200ms response latency and VAD-based natural turn detection.
How I work
Services
Three engagement models, all focused on production RAG for regulated industries. If your problem is outside this scope, I'll recommend someone better suited.
RAG Architecture Audit
For teams who have a RAG system in production but suspect it could be 2× more accurate. I audit your retrieval pipeline, chunking strategy, and prompt design, then deliver a prioritized roadmap.
Deliverables
- →Retrieval accuracy benchmark on your data
- →Architecture diagram of current vs. recommended pipeline
- →Prioritized fix list with effort estimates
- →2-hour walkthrough call with engineering team
Production RAG Build
For teams shipping a RAG system from scratch in a regulated industry. End-to-end build: document parsing, hybrid retrieval, citations, RBAC, observability. Production-ready, not a research demo.
Deliverables
- →Custom document parsers for your formats
- →Hybrid retrieval pipeline (vector + keyword)
- →FastAPI backend with async processing
- →Source citations with audit trails
- →Multi-tenant RBAC if applicable
- →Observability and eval pipeline
Fractional AI Architect
For founders or CTOs making AI architecture decisions but lacking in-house expertise. Async + 2 calls/month. Architecture reviews, vendor selection, hiring guidance, technical due diligence.
Deliverables
- →Async Slack/email access
- →2× 60-min calls per month
- →Architecture review documents
- →Hiring + vendor recommendations
- →Technical due diligence on AI claims
Background
About
I'm Muhammad Haseeb, an AI and Backend Engineer based in Islamabad, Pakistan. I specialize in production Agentic RAG for regulated industries: legal, energy, and government.
My path wasn't a straight line into AI. Every FastAPI service, every database migration, every ETL pipeline was prelude to the same realization: the hardest problem in production AI isn't the LLM. It's building systems where compliance and auditability aren't nice-to-haves. They're the architecture.
I build systems where every answer is cited, every retrieval is auditable, and the pipeline holds up under enterprise load.
Experience
Apr 2024 – Present
AI & Backend Engineer
Stixor Technologies
- •Designed Agentic RAG assistants processing 300+ GB of multi-format enterprise data (PDF, Word, Excel, PowerPoint, SAP), including 1K–6K page technical documents, for a major energy company.
- •Built the Malakah AI Legal Assistant achieving 98% recall@5 on Saudi regulations, contributing to $600K in pre-seed funding.
- •Architected a multi-tenant AI SaaS platform with RBAC, Qdrant vector search, and hybrid retrieval using PydanticAI.
- •Engineered a bilingual (English/Urdu) real-time voice agent on LiveKit with Whisper STT, ElevenLabs TTS, and VAD-based turn detection.
- •Engineered FastAPI inference backends supporting 1K+ RPS with async processing, caching, and connection pooling.
Dec 2023 – Feb 2024
Python Developer (Contract)
TriTech Solutions
- •Developed a FastAPI RESTful API for an e-commerce platform, integrating AWS DynamoDB for scalable catalog management.
- •Deployed services on AWS EC2 with production-grade configuration.
- •Introduced Pytest-driven unit testing and CI practices.
Mar 2023 – Apr 2024
Remote AI/ML Engineer
Freelance
- •Built a hybrid deep learning model for precipitation nowcasting using the CIKM radar dataset (RMSE = 11 dBZ).
- •Developed a real-time EfficientNet-based classifier for Pakistani currency notes achieving 99% accuracy.
- •Prototyped a multilingual voice-command eCommerce system using GCP Speech-to-Text and Translation APIs.
Available for new projects
Verified reviews from enterprise clients on Upwork.
What people say
Testimonials
I had the chance to oversee Muhammad across several AI initiatives, and he consistently delivered reliable, production grade work. His contributions to the Mari Energies Chatbot, especially the RAG pipeline and agentic architecture, showed real engineering maturity and a strong grasp of enterprise standards. He carried that same discipline into the Sky47 MultiTenant platform. Muhammad is thoughtful, dependable, and committed to building AI systems that work in the real world.
Dr. Hassan Sherwani
VP, AI & Data Platforms — managed Muhammad directly
Haseeb stood out as a highly motivated and intellectually curious student who consistently went beyond the curriculum. His ability to grasp advanced concepts... made him a standout. I have no doubt that Haseeb will continue to grow as a thoughtful and capable data scientist.
Muhammad Arshad
Associate Professor | Dean
Haseeb is an exceptional Python developer who stands out for his remarkable skills. I was particularly impressed by his ability to juggle multiple tasks with ease, great time management, and self sufficiency. He's dedicated, innovative, and a true asset to any team.
Azaz Rashid
Sr. Python Developer
I had the privilege of being mentored by Muhammad Haseeb... He provided invaluable guidance during and even after the program. Muhammad Haseeb also helped me through chats with the challenges I was facing. I learned a great deal from him.
Abdul Rehman
Software Engineer
Haseeb is an exceptional individual with a passion for community service... Not only is he a skilled public speaker, but he also possesses a deep understanding of Machine Learning. I highly recommend Haseeb for any opportunity that values community engagement and technical expertise.
Mahnoor Rana
AI Engineer
I had the chance to oversee Muhammad across several AI initiatives, and he consistently delivered reliable, production grade work. His contributions to the Mari Energies Chatbot, especially the RAG pipeline and agentic architecture, showed real engineering maturity and a strong grasp of enterprise standards. He carried that same discipline into the Sky47 MultiTenant platform. Muhammad is thoughtful, dependable, and committed to building AI systems that work in the real world.
Dr. Hassan Sherwani
VP, AI & Data Platforms — managed Muhammad directly
Haseeb stood out as a highly motivated and intellectually curious student who consistently went beyond the curriculum. His ability to grasp advanced concepts... made him a standout. I have no doubt that Haseeb will continue to grow as a thoughtful and capable data scientist.
Muhammad Arshad
Associate Professor | Dean
Haseeb is an exceptional Python developer who stands out for his remarkable skills. I was particularly impressed by his ability to juggle multiple tasks with ease, great time management, and self sufficiency. He's dedicated, innovative, and a true asset to any team.
Azaz Rashid
Sr. Python Developer
I had the privilege of being mentored by Muhammad Haseeb... He provided invaluable guidance during and even after the program. Muhammad Haseeb also helped me through chats with the challenges I was facing. I learned a great deal from him.
Abdul Rehman
Software Engineer
Haseeb is an exceptional individual with a passion for community service... Not only is he a skilled public speaker, but he also possesses a deep understanding of Machine Learning. I highly recommend Haseeb for any opportunity that values community engagement and technical expertise.
Mahnoor Rana
AI Engineer
Get in touch
Ready to build?
If you're building production AI for a regulated industry, or need an architecture review, let's talk. 15 minutes is enough to know if it's a fit.
Or email directly at mhaseeb1604@gmail.com