AI & ML Engineering
Designing and deploying machine learning systems that are measurable, maintainable and aligned with product goals.
- Model design & evaluation
- MLOps & monitoring
- Experimentation frameworks
AI Solutions Architect · Technical Delivery Leader
I’m Arpit Desai, I design systems that scale.
I help teams turn ambiguous product and technology ideas into reliable, production‑ready web applications. From architecture and solution design to hands‑on delivery, I care about real‑world impact, not just diagrams.
What I do
- Design and ship AI features end‑to‑end
- Evaluate and integrate LLMs into products
- Set up data and experimentation pipelines
- Partner with product & engineering teams
Experience at the intersection of AI & product
Over the past years I’ve worked across AI, data and engineering – leading initiatives from idea to production deployment, and helping teams adopt AI responsibly.
LLM & Generative AI
Building LLM‑powered workflows, agents and assistants that are safe, grounded and tuned for specific domains.
- Prompt & system design
- Retrieval & grounding
- Evaluation & guardrails
Selected client projects
A quick look at some of the web applications and platforms I’ve helped design and deliver as Architect, Technical Lead and Senior Engineer.
Enterprise reward points platform
Architecting and leading the front‑end for a large‑scale reward points program (Kona) with a 60+ member team for a US‑based client.
Factory Bucket – manufacturing product
Senior developer on a Canada‑based manufacturing platform using Vue.js and Laravel, working directly with clients on both front‑end and back‑end.
Appraisal & banking platform
Work on a high‑traffic US appraisal system and banking product serving millions of users, including multiple systems integrations.
AI solutions & contributions
In parallel to my general project work, I design and deliver AI‑driven solutions across generative AI, agentic systems and automation.
Knowledge assistant for internal teams
Solutioned and architected an internal assistant that connects tickets, documentation and tools so teams can answer complex questions quickly.
Agentic workflow orchestration
Designed agent‑based workflows that triage, enrich and route work automatically while keeping humans in control of approvals and critical decisions.
Evaluation & guardrail frameworks
Defined evaluation approaches for AI features, including metrics, datasets and guardrails so teams can ship with confidence instead of guesswork.
AI‑assisted developer tools
Worked on patterns and tools that help developers use AI for code review, refactoring suggestions and documentation, while keeping them in control.
Domain‑specific copilots
Designed domain‑aware AI copilots that combine structured data, business rules and LLMs to support users in specialised workflows.
Writing by Arpit Desai
I write about the reality of building AI products, lessons from shipping systems, and how I think about responsible AI.
Designing AI systems that survive production
A practical mindset for turning promising prototypes into robust AI features your team can operate confidently.
How I evaluate LLM‑powered products
A structured approach I use with teams to define success metrics, evaluation datasets and guardrails for LLM features.
Tell me what you want to build
Whether you’re exploring AI for the first time or scaling existing systems, I’d love to hear about your challenges and goals.
Contact Arpit Desai
Prefer email? You’ll find a direct way to reach me on the contact page.