I’m an independent AI consultant and engineer focused on building reliable LLM applications and the cloud infrastructure required to operate them.
I work across the full path from business problem to production system: retrieval, agents, evaluation, observability, data pipelines, CI/CD, cloud deployment, security, and platform operations.
- Knowledge-heavy workflows that require accurate retrieval and citations
- AI agents that use tools and coordinate multi-step work
- Evaluation systems for non-deterministic model behavior
- Production observability for latency, cost, quality, and failures
- Data pipelines for documents, context, and evaluation datasets
- Secure MCP and agent workflows
- Human-in-the-loop systems for ambiguous or consequential decisions
I’m interested in work involving:
- Product AI engineering
- Forward-deployed AI systems
- AI platform and reliability
- Retrieval and evaluation
- Agentic workflows
- Secure enterprise AI
- Production data and model infrastructure
If you need an engineer who can move from business problem to working system, then measure what happened and explain why, I’d be glad to talk.
LinkedIn: https://www.linkedin.com/in/sandipan-haldar/
Website: https://www.sandipanhaldar.com/
Email : work@sandipanhaldar.com




