Build production AI that actually ships.

You'll lead the design and delivery of agentic AI, RAG, and generative systems for our most strategic client engagements — the BIBA-style projects where a single well-architected pipeline decides whether AI ships or stalls. You'll be the technical lead in the room when we scope, and the engineer in the IDE when we build.

• Own the architecture of multi-step agent and RAG systems end-to-end, from data ingest through eval
• Pick the right model and orchestration pattern — including when *not* to use an LLM
• Set up evaluation harnesses and observability on day one so 'is this better?' has an answer
• Embed with client teams; lead workshops; drive technical decisions with conviction
• Mentor mid-level engineers and review architecture across other engagements

• You've shipped LLM-powered systems to real users — not demos, not internal tools
• You've debugged a RAG system that confidently cited the wrong document and fixed it for real
• You can argue both sides of 'fine-tune vs. prompt' and know when each is right
• You read papers but you ship code; you read the OpenAI/Anthropic changelogs and try things
• You're comfortable owning a client conversation, not just an issue ticket

You'll own end-to-end product builds — React / Next.js on the frontend, Node and Python on the backend — for AI-powered platforms that go straight into production. The work is concrete: ship features people use Monday morning, instrument them so we know they're working, iterate fast.

• Own product builds from architecture through ship — frontend, backend, infrastructure
• Design APIs and data models that hold up under real traffic and real audit
• Integrate LLM and ML features into product surfaces with eval and feature flags
• Build for compliance when the engagement demands it — HIPAA, SOC 2, GDPR-ready patterns
• Set the pace for code quality, code review, and engineering practice on the team

• You've shipped Next.js (or similar) apps with non-trivial backend state, not toy demos
• You enjoy the seams — auth, observability, deploy pipelines — as much as the features
• You've added AI to an existing product and felt the difference between a demo and production
• You know when to reach for Postgres and when to reach for the warehouse
• You'd rather ship a smaller working thing than a bigger half-finished one

You'll build cross-platform mobile experiences for AI-powered products — telehealth flows, field tools for agritech, retail merchandiser apps. The constraint is interesting: on-device performance, offline-first when needed, accessibility-first always.

• Build cross-platform mobile apps (React Native, Flutter, or native iOS/Android as fits)
• Integrate AI features that respect battery, network, and latency budgets on real devices
• Drive accessibility, motion, and gesture design that holds up under real-world usage
• Own the store release process — signing, review prep, phased rollout, crash monitoring

• You've shipped apps to the App Store and Play Store and lived through review cycles
• You know when to use a WebView and when to spend the week on native
• You've integrated on-device or remote AI features and instrumented the experience honestly
• You care about the 200ms feel, not just the feature list

You'll build the data plumbing AI products need — ingestion, transformation, warehousing — plus the PowerBI / Microsoft Fabric dashboards on top. The Khaitan Education engagement (paper generation + warehouse + PowerBI) is a flagship of this work; you'll lead the next one.

• Design and ship multi-source data pipelines with CDC, transformation orchestration, and lineage
• Model warehouses for AI workloads — feature stores, vector stores, analytical surfaces
• Implement PowerBI / Microsoft Fabric dashboards that decision-makers actually open
• Set up observability and data-quality monitoring so 'is the data fresh?' has an answer

• You've built warehouses on Postgres, Snowflake, BigQuery, or Fabric — and shipped them past audit
• You're fluent in PowerBI / DAX / Power Query, and have opinions about M-code
• You've designed CDC pipelines that didn't break the source database
• You enjoy translating fuzzy business asks into clean, queryable models

You'll own the deploy pipelines, infrastructure-as-code, and observability that keep our client systems alive. The kind of work where 'invisible' is the highest praise — when production stays up, the deploy ships at 5 PM Friday, and the alerts make sense at 2 AM.

• Own CI/CD, infra-as-code (Terraform / Pulumi), and container orchestration across engagements
• Set up observability, alerting, and on-call patterns the team can actually use
• Drive compliance posture — HIPAA, SOC 2, ISO 27001 — into the platform layer
• Optimize cloud spend without sacrificing reliability or velocity

• You've run production AI workloads on AWS, GCP, or Azure at non-trivial scale
• You write Terraform like prose and document runbooks like you've been paged before
• You believe 'observability isn't a tool, it's a habit' and live by it
• You can have a frank conversation with a CTO about cloud costs