Symbian Health
Software Engineer
USA · Remote
  • Owned the end-to-end EMR-to-reporting pipeline - consistent, durable, audit-ready clinical workflows at production scale.
  • Shipped a multi-agent LLM orchestration system automating patient-priority scheduling and hospital task routing.
  • Built a RAG-based clinical context engine on Amazon Bedrock - retrieving relevant patient history per visit and generating concise doctor-ready summaries.
  • Built a real-time patient behaviour monitoring system detecting fall risk, respiratory distress, and onset of illness before adverse events occur.
LLMsMulti-Agent RAGAmazon Bedrock AWS CDKPythonHIPAA
DRDO — CABS
Junior Research Fellow
Bangalore, India
  • (Details limited due to DRDO confidentiality policies.)
  • Implemented state-of-the-art algorithms for angle estimation and adaptive jamming suppression for operational airborne radar systems.
  • Architected a high-performance monolithic system with distributed compute for deterministic real-time radar signal processing.
  • Engineered a real-time pre-emptive EDF scheduler on bare-metal hardware and Fedora Linux for hard-deadline radar workloads.
Signal ProcessingC / C++ CUDAIntel MKL BLAS/LAPACKReal-Time OS
Rovlin
AI Advertisement & Outreach Platform
Live Product
  • Built and launched an AI-native outreach platform that leverages LLMs to generate hyper-personalised advertisement copy and automate multi-channel campaigns at scale.
  • Designed a multi-agent AI pipeline that autonomously researches targets, crafts contextual messaging, and optimises delivery timing to maximise engagement.
  • Architected the full platform - from AI inference layer to production-grade APIs - handling end-to-end campaign lifecycle management for real users.
LLMsMulti-Agent NLPAWS Live www.rovlin.com ↗
Hallucination Detection
AI Trustworthiness Assessment Pipeline
Open Source
  • Built a rigorous pipeline to detect LLM hallucinations using token log-probabilities and semantic similarity, with multi-stage factual verification.
  • Implemented cross-reference checks, statistical scoring, and factual consistency modelling to systematically quantify model reliability.
  • Delivered a real-time 0-100 confidence scoring system providing interpretable trustworthiness signals for any LLM response.
NLPLog-Probs PythonHuggingFace Open Source GitHub ↗
AI / ML
LLMsRAG AgentsPyTorch HuggingFaceLangChain BedrockNLP
Cloud
AWS CDKDocker ServerlessCI/CD
HPC
CUDAIntel MKL BLAS/LAPACKGPU
Languages
PythonC C++TypeScript
Web & APIs
FastAPIDjango ReactNode.js
Healthcare
HIPAADICOM ICD-10STT/TTS

Let's work together

Open to full-time roles, freelance, and interesting collaborations in AI and systems engineering.

✉  satviksingh1000@gmail.com