AI engineer focused on clinical LLM systems, healthcare interoperability, and applied biometric ML.
I build production-oriented healthcare AI: clinical summarization, medical entity extraction, ontology mapping, FHIR/HL7 workflows, and multi-agent systems. I am currently working as an LLM and multi-agent developer at ArogyaPandit, with research interests across clinical NLP, causal reasoning, privacy-aware medical AI, and biometrics.
Currently building MediLipi: doctor-patient conversations -> Whisper ASR -> medical NER -> SNOMED CT / UMLS / MedDRA mapping -> causal Bayesian networks for structured clinical knowledge.
- Best fit: AI/ML Engineer, Clinical NLP Engineer, Healthcare AI Engineer, Applied Research Engineer.
- Core strengths: LLM apps, structured extraction, RAG, hallucination guardrails, FHIR/HL7, medical ontologies, multi-agent workflows.
- Research signal: International competition work across bioacoustics, face, sclera, and ear biometrics.
- Open to: healthcare AI, clinical NLP, biometrics, and agentic AI roles.
| Project | What it demonstrates | Stack |
|---|---|---|
| crossroad-fhir-link | HL7/FHIR IPS demo converting diabetes reports into coded FHIR bundles, terminology evidence, receiver readiness checks, and country-specific PDFs. | TypeScript, FHIR R4, Vercel |
| MediScribe | Clinical scribe prototype for recording, transcribing, summarizing, and storing patient-clinician encounters. | Python, notebooks, healthcare NLP |
| mosquito123 | BioDCASE 2026 cross-domain mosquito species classification with domain-balanced audio ML training. | Python, PyTorch, audio ML |
| Jalguard-final | Public health command-center app concept for water-borne disease monitoring, offline field workflows, and role-based dashboards. | Flutter, Dart, SQLite |
| Area | Signal |
|---|---|
| Bioacoustics | BioDCASE 2026 mosquito species recognition work focused on unseen-domain generalization. |
| Biometrics | IJCB competition work spanning face liveness, morphing attack detection, sclera segmentation, face recognition, and ear biometrics. |
| Clinical reasoning | MediLipi research direction combining medical ontologies with causal Bayesian networks. |
- Languages: Python, TypeScript, JavaScript, Dart, SQL
- AI/ML: PyTorch, Transformers, spaCy, Whisper, Llama, Gemini, OpenAI APIs, Ollama
- Healthcare: FHIR R4, HL7, IPS, SNOMED CT, UMLS, MedDRA, clinical summarization, medical NER
- Agents/Data: LangChain, MCP, RAG, structured outputs, Pydantic, DoWhy, pgmpy
- Delivery: Docker, GCP, Vercel, GitHub Actions, Linux
- Clinical AI systems that turn messy notes and conversations into reliable structured data.
- Interoperability tools that make healthcare records portable across systems and countries.
- Multi-agent workflows where each agent has a clear job, evidence trail, and failure mode.
- Applied ML prototypes that move from notebook experiments to usable demos.
- Email: anantsushilgupta@gmail.com
- LinkedIn: linkedin.com/in/anantgp
- GitHub: github.com/AnantGp