FAQ
Frequently asked questions
Everything you might want to know about working with me — process, pricing, time zones, and how I build. Still have a question? Just ask.
Working together
What does Shahzad Malik do?
I'm a freelance data scientist and AI engineer. I help startups and enterprises with event tracking and analytics, production machine learning and MLOps, LLM and RAG applications, A/B testing and experimentation, AI workflow automation, and data visualization — end to end, from instrumentation to production.
Where are you based and what time zones do you cover?
I'm based in Islamabad, Pakistan (PKT) and work fully remotely. That gives me solid working-hours overlap with both European and US time zones. Engagements are run async-first, so progress never blocks on a single daily call.
Do you work with international clients?
Yes — I work with clients worldwide across the US, Europe, and Asia, fully remotely.
Engagement & process
How do engagements start?
With a free 45-minute discovery call to map your data stack, goals, and the fastest path to ROI. I then send a scoped proposal with deliverables, timeline, and transparent pricing. If it's a fit, we start with weekly check-ins and working software each sprint.
How do you price projects?
Most work is fixed-price against clearly scoped deliverables, so you know the cost up front. Time-and-materials is available for open-ended or exploratory work. Pricing depends on scope and complexity, which we nail down in the proposal — no retainer is required to start.
What happens after the project ships?
Every engagement ends with full documentation, team training, and a 30-day window of included support, so your team can own and run what we built confidently.
Technical
Can you work within our existing stack?
Yes. I target your current cloud (AWS, GCP) and tools rather than forcing a migration, using managed services where they save time and custom infrastructure only where it pays off.
Do you build LLM features that are safe for production?
Yes — I ground answers in retrieval, require citations, build evaluation harnesses that catch regressions before release, and add guardrails so features degrade gracefully instead of hallucinating. See the LLM / RAG service page for details.
Can you fix or productionize something we already started?
Often yes. Common engagements include cleaning up a messy analytics setup, productionizing an existing model, or hardening a RAG prototype with evals and guardrails.