Amit Kumar S.

I help healthcare teams turn operational problems into working AI solutions from understanding clinical workflows and defining the right use case to building, integrating and supporting the system. I work as a Forward Deployed AI Engineer: collaborating directly with founders, clinicians and operational teams, translating their requirements into technical decisions, and staying hands-on through implementation. My background includes 10 years in software development and 3.5+ years working with AI agents, RAG and LLM applications. On Upwork: $50K+ earned | 100% Job Success | Top Rated Plus WHERE I CAN HELP • Identify where AI can reduce manual work and where a simpler automation is sufficient. • Turn an early idea into a scoped pilot with clear deliverables and acceptance criteria. • Build healthcare AI agents, document assistants and doctor-facing voice workflows. • Integrate AI into existing applications, APIs, databases and operational processes. • Investigate retrieval failures, unreliable agent behaviour, latency and infrastructure costs. • Help teams move from a working demonstration to a deployable, maintainable system. RELEVANT FIELD EXPERIENCE University-led cancer support project — ongoing contract Developing a conversational support system for breast cancer patients and caregivers. My work connects support requirements with specialist-agent routing, conversational memory, guideline retrieval and backend integration, including safety checks and human escalation paths. Susrut Eye Foundation — doctor voice assistant Worked on a doctor-facing voice assistant using Sarvam and OpenAI, supporting clinical documentation and prescription-drafting workflows from spoken input. Helix Health — healthcare communication Took ownership of calling-module reliability issues in an existing healthcare application. Improved incoming-call handling, duplicate-call prevention and connection performance using React Native and LiveKit. The client awarded a five-star review, highlighting ownership, problem-solving and improvements to the user experience. EdTech AI platform — RAG optimisation Worked on retrieval and response tuning in an existing AI platform, receiving a five-star client review. My broader experience includes legal document intelligence, large-scale document indexing and full-stack product development. HOW I APPROACH AN ENGAGEMENT Understand the workflow Identify the users, operational bottleneck, existing systems, data constraints and business objective. Define a practical first milestone Agree on scope, success measures and acceptance criteria. Make the trade-offs between accuracy, latency, cost and implementation effort explicit. Build and validate Implement the workflow, connect the required systems and test against representative examples. Use stakeholder feedback to refine the solution. Deploy and hand over Address failure handling, access controls, sensitive-data handling, logging and monitoring. Provide documentation so the team can operate and extend the system. TECHNICAL FOUNDATION Python, FastAPI, LangGraph, LangChain, OpenAI, Claude, Amazon Bedrock, Sarvam, PostgreSQL/pgvector, AWS and Docker. For product integration: React, Next.js, React Native, LiveKit and WebRTC. Tell me which workflow is slowing your team down, what systems you already use and what a successful outcome would look like. I’ll help translate that into a practical implementation plan and a working first milestone.