I'm Hassan Shahzad
Agentic AI Engineer
Agentic AI that pays for itself. I design, ship, and industrialise end-to-end systems built for measurable business outcomes.
5+ years across enterprise GenAI, MLOps, and data engineering. Currently at SAP Signavio while completing an M.Sc. Data Science at FAU.
5+ years
AI & ML engineering
Agentic AI · GenAI
Multi-agent · RAG · MLOps
M.Sc. Data Science
FAU · in progress
Nuremberg, Germany
Open to opportunities
About Me
Who I am, and how I work
I build AI systems that ship and move real numbers.
I'm an Agentic AI Engineer with 5+ years designing, industrialising, and deploying end-to-end AI systems in Python. My work spans multi-agent workflows, retrieval-augmented generation, NLP applications, and enterprise GenAI platforms. The goal is always the same. Build AI that delivers real outcomes, not another expensive prototype that never leaves the lab.
I take technical ownership from solution design through architecture, backend development, AI evaluation, cloud deployment, data pipelines, and MLOps. Today I develop secure enterprise GenAI applications at SAP Signavio while completing an M.Sc. Data Science at FAU. I care less about model novelty and more about systems that deliver measurable results in production.
Value Before Novelty
I start from the business outcome, not the model. Every system I build has a number attached to it, whether that is cost saved, revenue added, or hours returned to a team.
The Whole System, Not the Model
I own the full path from solution design to architecture, backend, cloud deployment, and data pipelines. Models are the easy part. Shipping them into production is where I live.
Reliable in Production
I treat evaluation and MLOps as first-class work with Docker, CI/CD, MLflow, and LangSmith. AI that cannot be trusted release over release is not finished, it is a demo.
Companies don't need another AI demo. They need systems that generate value in production. That's what I build.
Selected Impact
Results, not just prototypes
Real systems, real numbers. Every figure here comes from work that shipped and generated value for the team that ran it.
Annualised operational savings
Agentic sales-automation platform at DTC Force
Annualised cost savings
LLM platform for BRDs and Salesforce assets
Support cost reduction
Convoi AI voice agent at Venture AI
Weekly sales uplift
Recommender system at Bazaar (+15% AOV)
GMV-loss reduction
Demand forecasting across 6,000+ items
Fraud losses cut
Fraud-detection system at Bykea
Experience
Where I've delivered
Seven roles across enterprise GenAI, AI engineering, and data science.
AI Engineer, Working Student
SAP Signavio · Munich, Germany
- Deployed a secure internal alternative to NotebookLM on SAP BTP (SAP AI Core + XSUAA), enabling 25 to 30 product-team members to ingest transcripts, run semantic search, and generate grounded insights.
- Built a multi-agent AI copilot on Signavio Process Insights (PINS) data for conversational process diagnostics and recommendations.
- Established an AI evaluation framework with manual qualitative checks plus a reusable offline Q&A test set to gauge response quality after each release.
AI Engineer
DTC Force · Remote
- Designed and deployed an agentic sales-automation platform (lead sourcing and initial outreach) that delivered an estimated USD 200k in annualised operational savings.
- Built an LLM platform generating BRDs and Salesforce assets (LWC, Apex, metadata) for an estimated USD 150k in annualised cost savings.
- Engineered modular Python and FastAPI services (Apollo, LinkedIn, email), containerised with Docker, automated CI/CD via GitHub Actions, and traced agent execution with LangSmith.
Lead AI Engineer
Venture AI · Lahore, Pakistan
- Led a cross-functional team of 5 delivering Convoi AI, an AI voice agent (Twilio, ElevenLabs, OpenAI, AssemblyAI) that reduced support costs by USD 10k per month.
- Built a voice product recommender combining RAG similarity retrieval with live PostgreSQL inventory queries, cached recent-item retrieval, and a cold-start strategy.
Senior Data Scientist
Bazaar Technologies · Islamabad, Pakistan
- Designed a collaborative-filtering recommender on behavioural and geolocation data that increased weekly sales by PKR 8M and AOV by 15%, validated via A/B testing.
- Built time-series demand forecasting for 6,000+ items that reduced gross-merchandise-value loss by 45%.
- Implemented model lifecycle management with MLflow, built fraud pipelines on AWS, and ran ETL on Airflow and GCP.
Data Analyst
Bykea · Islamabad, Pakistan
- Developed an XGBoost fraud-detection model that reduced fraud losses by PKR 100k per month.
- Tested driver bonus-payout scenarios that increased ride fulfilment from 55% to 73%.
Junior Data Scientist
Data Pilot · Lahore, Pakistan
- Built and deployed a Random Forest model on advertising data to predict marketing-campaign objectives and cut inefficient ad spend for a fashion brand.
Data Analytics Intern
Telenor · Islamabad, Pakistan
- Built an ETL pipeline consolidating multiple sources plus a Looker Studio dashboard, and was selected as a Telenor Super Intern for the project.
Selected Projects
Systems I've shipped
AI agents, GenAI analytics, voice platforms, and RAG. Real, linkable work.
Voice AI · Customer Support & SalesConvoi AI
Problem
Support and sales teams needed human-like phone coverage without scaling headcount or sacrificing latency.
What I built
A voice AI platform for low-latency, human-like phone calls that handle customer support, outbound sales, and scheduling workflows end to end.
Impact
Automated phone workflows across support, sales, and scheduling.
Voice AI · E-commerceCartly · AI Voice Shopping Assistant
Problem
Online shoppers drop off during search and checkout when the experience is friction-heavy and impersonal.
What I built
A voice-first shopping assistant for e-commerce: voice product search, personalized recommendations, quick add-to-cart, and voice checkout.
Impact
Voice-first product discovery and checkout for e-commerce.
Agentic AI · ChatbotSalesforce Assistant
Problem
Salesforce implementations stall on slow, manual requirements gathering and BRD authoring.
What I built
A chatbot that generates detailed BRDs and drives Salesforce solution design with AI-assisted precision.
Impact
Accelerated Salesforce requirements and solution design.
GenAI · Analytics PlatformSales Call Analysis Platform
Problem
Sales orgs had thousands of recorded calls but no scalable way to extract summaries, sentiment, and coaching insights.
What I built
An end-to-end platform that extracts summaries, sentiment, and insights from sales calls via an AWS/BigQuery pipeline surfaced in Looker Studio dashboards.
Impact
Turned raw call recordings into structured, queryable sales insight.
Agentic AI · AutomationExecutive AI Assistant
Problem
C-level users lose hours daily to email triage, calendar coordination, and messaging across disconnected tools.
What I built
A LangGraph-based executive assistant that manages email, calendar, and WhatsApp workflows with tool integrations and guardrails.
Impact
Automated email, calendar, and messaging workflows for executives.
GenAI · Retrieval-Augmented GenerationRAG Natural-Language-to-SQL
Problem
Non-technical stakeholders couldn't query databases directly, bottlenecking every data question on engineers.
What I built
A RAG system that translates natural-language questions into SQL over PostgreSQL, using Azure OpenAI for generation and Pinecone for schema/context retrieval.
Impact
Enabled self-serve database querying in plain English.
Want something similar for your business? Let's discuss your use case.
Let's TalkTechnical Expertise
The stack behind the systems
The stack I use to take AI from idea to production value, agentic systems first.
Agentic AI & GenAI
Software Development
Data Science
Data Engineering
Cloud & Platforms
MLOps / DevOps
Credentials, Education & Languages
Credentials
- Databricks ML Practitioner · Certified
- DataCamp Associate Data Scientist · Certified
- Udemy Deep Learning · Certified
- Telenor Super Intern · Selected for project impact
Education
- M.Sc. Data Science · Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)
- B.Sc. Computer Science · National University of Sciences and Technology (NUST)Sep 2017 to Jun 2021
Languages
- English · C1
- German · A2
Contact
Let's build it together
Hiring for an AI/ML role, or need an AI system for your business? Tell me about it. I reply personally.
Based in Nuremberg, Germany · hassanshahzadkhan3@gmail.com