About
Ankur Sharma, PhD — Computational biologist and agentic AI engineer, Singapore.
I build systems that connect genomic data to real-world decisions: outbreak surveillance, precision diagnostics, and autonomous research pipelines.
Full portfolio: ankurgenomics.github.io/agentic-genomics/
Background
- PhD in computational biology / bioinformatics, NTU Singapore
- 8+ years building production ML and genomic data pipelines in clinical and research settings
- Current focus: agentic AI systems for high-stakes biological decisions
Projects
- Reviewer2 — Autonomous ACMG/AMP variant second reviewer. 4-node LangGraph pipeline exposed as an MCP server. 21 tests, 86% action-band concordance vs expert-panel ClinVar.
- sentinel-amr — Explainable AMR classifier for bacterial surveillance. XGBoost + SHAP + LangGraph. ROC-AUC 0.894 on a group-aware holdout. Detects novel organisms and escalates before trusting the model.
- outbreak-agent — LangGraph triage pipeline for infectious disease surveillance. 33 tests, no API key required.
- agentic-genomics — Reasoning-traceable variant interpretation agent. VCF + HPO → ranked ACMG candidates with full audit trail.
- genomics-skills — 8 agent-callable genomics tools: TCGA expression, survival, enrichment, PubMed, protein mapping.
Contact
- LinkedIn: linkedin.com/in/ankurit
- GitHub: github.com/ankurgenomics
- Email: ankurs103@gmail.com