Open to relevant roles globally

Agentic AI & ML Engineer building reasoning systems for science.

PhD with 8+ years shipping production ML and cloud infrastructure across clinical diagnostics and genomics. I build multi-agent systems that diagnose real pipeline failures, ML-powered workflows that process 6,000+ samples, and open-source tooling for agentic AI in biomedicine. Open to relevant opportunities globally — across industry, research, and startups.

8+ yrs
Production ML & bioinformatics
6,000+
Genomic samples processed in production
5 agents
Multi-agent ops system shipped (GenomicsOps AI)
40% ↓
Compute cost savings on production pipelines
// Selected work

Agents, pipelines, and research projects

Projects spanning genomic foundation-model fine-tuning, open-source agentic AI for variant interpretation and outbreak surveillance, agent-callable genomics tools, production multi-agent systems, self-optimising ML pipelines, and applied research. Each links to code, a demo, or a writeup where possible.

Open source · Explainable ML XGBoost · SHAP · LangGraph · Python

sentinel-amr — Explainable AMR Classifier

Explainable, agentic antimicrobial resistance classifier for bacterial pathogen surveillance. Predicts resistance, names the genes that drove the call, detects organisms outside the training distribution, and escalates uncertainty to a human.

Trained on 298 real K. pneumoniae genomes from BV-BRC (PATRIC), labelled by laboratory AST. A 5-node bounded LangGraph agent runs vectorisation → XGBoost classification → alignment-free k-NN novelty check → SHAP TreeExplainer attribution → critic gate. The critic never loops — it is a single deterministic pass. Every run produces an auditable markdown report (for a risk officer) and a JSON record (for a LIMS). Top SHAP drivers independently rediscovered correct biology: CTX-M-15, acrAB efflux pump, DNA topoisomerase IV — the literal fluoroquinolone target. Novel-organism flag blocks the automated call when Jaccard distance exceeds the calibrated threshold; the model's probability is explicitly not trusted.

0.894
ROC-AUC, group-aware holdout
298 genomes
real BV-BRC AST labels
8 tests
dataset, split, novelty, agent
XGBoost SHAP TreeExplainer LangGraph BV-BRC / CARD / VFDB GroupShuffleSplit MIT
Open source · Agentic AI LangGraph · Python · No API key

outbreak-agent -- Infectious Disease Triage Pipeline

4-node LangGraph state machine that triages infectious disease cases -- built around the April 2026 MV Hondius / Andes virus event, the first confirmed human-to-human hantavirus transmission on a cruise ship.

A self-correcting agentic pipeline: genomic_node identifies clade and mutations, linkage_node resolves contact clusters and transmission mode, risk_node scores 0-100 with tier (LOW / MEDIUM / HIGH / CRITICAL), and critic_node enforces four consistency rules -- looping back up to 3 times if flags are raised. Generates a 3-panel matplotlib risk dashboard (PNG) and a structured A4 PDF triage report automatically on every run. No API key. No cost. 33 tests, fully deterministic. Part of the same ecosystem as gwas_nf: gwas_bridge.py feeds REGENIE tophits directly into this triage pipeline for population-to-clinical interpretation.

4 nodes
genomic / linkage / risk / critic
33 tests
23 unit + 10 integration, free to run
<2 sec
CRITICAL triage, MV Hondius case
LangGraph 0.6 LangChain matplotlib ReportLab pytest Apache 2.0
Open source · Agent skills Python · Claude Haiku · REST APIs

genomics-skills — Agent-Callable Skill Library

8 pure-Python genomics skills that downstream agents can call: expression profiling, survival analysis, protein mapping, pathway enrichment, literature search, and more.

The downstream skill layer for agentic-genomics. Each skill is a standalone, agent-discoverable Python module with a SKILL.md contract, CLI entrypoint, and deterministic output (TSV + PNG/SVG). Pan-cancer expression uses real TCGA data (9,479 samples across 31 cancer types via cBioPortal). Kaplan-Meier survival runs Cox PH regression on actual patient data. LLM-powered routing via Claude Haiku maps natural-language queries to the right skill. Parquet caching makes repeat queries instant.

8 skills
Agent-callable, SKILL.md contract each
9,479
Real TCGA patient samples
MIT
Open source, Python 3.9+
Python Claude Haiku cBioPortal API MyVariant.info NCBI E-utils PDB / AlphaFold Pandas Matplotlib

GenomicsOps AI

Five specialized agents that triage and resolve DRAGEN, ICA and SGE/HPC pipeline failures end-to-end.

Personal project built on weekends: Trigger → Log Fetcher → RAG → Classifier → JIRA Writer. Tested on real failure scenarios — BED-file overlaps, samplesheet index mismatches, stuck SGE jobs — with end-to-end triage and ticket creation.

5
Specialised agents
Multi-agent Claude API RAG Python JIRA & Confluence APIs
Open source · Production · Cloud Nextflow · REGENIE · AWS

gwas_nf — Multi-Ethnic GWAS Pipeline

REGENIE-based Nextflow GWAS pipeline for the TEMUS multi-ethnic cohort (4 groups, 10,000 samples, 100,000 variants, 13 phenotypes) -- with downstream agentic interpretation via GenomicsCopilot.

End-to-end population genomics pipeline: genotype QC and LD pruning, REGENIE whole-genome regression (Step 1 + Step 2), multi-ethnic stratified association, interactive Manhattan plots per phenotype per group, and automated HTML reports. Top hits feed into GenomicsCopilot via gwas_bridge.py: variant IDs extracted from .regenie.filtered.gz tophit files and passed to the 7-node LangGraph interpretation pipeline for ACMG classification and LLM-powered clinical synthesis. Population discovery to clinical interpretation in a single command.

4 groups
TEMUS multi-ethnic cohort
13 phenotypes
Parallel GWAS runs
100k variants
REGENIE whole-genome regression
Nextflow DSL2 REGENIE AWS Batch Python R / ggplot2 Docker
Research · PhD

Age-dependent hepatocyte epigenomics

Integrative RNA-seq + ChIP-seq + Hi-C analysis pipeline revealing age-driven chromatin reorganisation in mouse liver.

PhD work at NTU: built end-to-end NGS analysis pipelines for transcriptome, histone modifications, and 3D chromatin. Identified H3K27me3 as a key age-dependent regulator. The technical stack — reproducible pipelines, multi-omic integration, careful statistics — is the same foundation I now use for agentic ML systems.

RNA-seq ChIP-seq Hi-C (3C-seq) R · Bioconductor Python k-means / GSEA
// About

Scientist-turned-engineer with production miles.

PhD bioinformatician with 8+ years spanning clinical diagnostics, genomic data analysis, and cloud infrastructure. Proven track record in regulatory validation (DVT, clinical concordance, LOD), production-grade pipelines, and commercial product improvement at leading genomics companies. Now focused on agentic AI systems that apply the same rigour — traceability, reproducibility, measurable outcomes — to autonomous reasoning over real biomedical data.

// Experience

From benchtop to agentic pipelines

A decade-plus moving from wet-lab genomics and pharmaceutical QC to production ML infrastructure and, now, agentic AI systems.

Jul 2025 — Present
Senior Scientist, Bioinformatics
Illumina · Singapore
  • Lead Design Verification Testing (DVT) and regulatory validation for commercial diagnostic products (TSO500, NIPT16, VeriSeq) — FDA compliance, clinical-deployment quality.
  • TSO500: optimised TruSight Oncology 500 assay workflows for comprehensive genomic profiling in oncology.
  • NIPT16: contributed to non-invasive prenatal testing product development and validation.
  • VeriSeq: performed software testing and workflow optimisation, improving reporting efficiency and data-analysis reliability.
  • Designed clinical concordance studies — sensitivity, specificity, accuracy, LOD, reproducibility — for regulatory submissions.
Jan 2022 — Jul 2025
Senior Scientist — Bioinformatics & Cloud
Mirxes · Singapore
  • Built and validated production-grade cloud data-analysis pipelines: 40%↓ compute cost, 50%↓ storage cost, 30%↓ turnaround time.
  • Analysed 6,000+ genomic samples with standardised workflows and rigorous QC.
  • Architected AWS infrastructure integrating 15 services to manage and process 400 TB of genomic data.
  • Led a team of 5 scientists delivering clinical diagnostic assay workflows across cancer types — 33%↓ analysis time.
  • Technical lead for the Singapore National Precision Medicine project; identified business opportunities worth SGD 0.5M.
Jul 2021 — Dec 2021
Scientist, Assay Development
Vela Diagnostics · Singapore
Implemented automated verification procedures and maintained regulatory documentation (NCR, CAPA, DR). Shipped a testing framework that cut non-conformance reports by 30% and lifted product quality by 15%.
2016 — 2021
PhD Research Fellow
Nanyang Technological University · Singapore
Multi-omic study of age-dependent transcriptional and epigenetic changes in mouse hepatocytes. Built reproducible NGS analysis pipelines covering RNA-seq, ChIP-seq, and Hi-C (3C-seq). Head of NTU 3MT team (23 members, Nanyang Awards); led operations for TEDxNTU (78-person team, 1,500+ audience). Thesis DOI: 10.32657/10356/155390.
2011 — 2013
Technical Supervisor
Zydus Cadila Healthcare Ltd. · Ahmedabad, India
Built process-validation and QC pipelines for 13 pharmaceutical products under cGMP guidelines — 30%↑ efficiency, 10%↓ processing time. First exposure to regulated, high-stakes data infrastructure.
Education
PhD · M.Tech · B.Tech — Biological Sciences & Biotechnology
NTU Singapore · UPTU India
  • PhD — Nanyang Technological University, Singapore (2016 – 2021).
  • M.Tech Biotechnology (2013 – 2015).
  • B.Tech in Biotechnology — Uttar Pradesh Technical University, India (2007 – 2011).
// Publications & writing

Research output & technical writing

Peer-reviewed work, open-source documentation, and engineering writeups.

Technical blog · Amazon Web Services · 2025
Using serverless for cross-organizations information exchange in genomic analysis
Published on the AWS Builder Center. Covers serverless architecture patterns for sharing genomic data and analysis workflows across organizational boundaries at scale. · Read on AWS
Technical writeup · Open source
Why agentic AI for genomics? Designing reasoning-traceable variant interpretation
Design philosophy & architecture doc shipped with agentic-genomics. · Read on GitHub
Doctoral thesis · NTU · 2021
Age-dependent transcriptional and epigenetic alterations in mouse hepatocytes
Sharma, A. (2021). Nanyang Technological University. · doi:10.32657/10356/155390 · hdl.handle.net
Conference poster · Cell Symposia, Chicago · 2019
Significance of hepatocyte polyploidization in liver physiology and pathology
Sharma A, Ong A, Wuestefeld T, Sanyal A. Transcriptional regulation in evolution, development and disease.
Peer-reviewed · Frontiers in Microbiology · 2018
Antiproliferative and antioxidative bioactive compounds in extracts of marine-derived endophytic fungus
Kumari M, Taritla S, Sharma A, Jayabaskaran C. · Frontiers in Microbiology
// Skills

What I use, how often

Weighted by how regularly I use each tool in production, not by how much I've ever touched it.

Agentic AI & LLMs primary

LangGraph / LangChaincore
Claude / Anthropic APIcore
Multi-agent orchestrationcore
RAG & vector storesfrequent
Tool-calling & function agentsfrequent
OpenAI API / GPT-4frequent
Local LLMs (Ollama)familiar

Machine learning & MLOps primary

Python · scikit-learncore
Pandas / NumPycore
Statistical modellingcore
PyTorchfrequent
Feature engineeringfrequent
MLflow / experiment trackingfrequent

Cloud & infra primary

AWS (Batch, Lambda, S3, Step Fns)core
Nextflow (DSL2)core
Dockercore
CI/CD (GitHub Actions)frequent
IaC (CloudFormation / CDK)frequent
Kubernetesfamiliar

Bioinformatics domain expert

NGS — WGS / WES / RNA-seq / ChIP-seqcore
WGMS · single-cell · spatialfrequent
DRAGEN · GATK · samtools · bcftoolscore
Variant calling & interpretationcore
R · Bioconductor · DESeq2frequent
Snakemakefrequent

Regulatory & clinical expert

Design Verification Testing (DVT)core
Clinical validation (LOD, sensitivity, specificity)core
FDA compliance · SOPscore
cGMP documentationfrequent
NCR / CAPA / DRfrequent
HIPAA · security compliancefrequent
// Contact

Let's build something useful.

Open to relevant roles globally — across industry, research, and startups — where agentic AI, ML, or computational biology intersects with real-world impact. Based in Singapore (PR) — open to remote, hybrid, or relocation anywhere in the world.

Based in Singapore (PR) · Open to remote, hybrid, or relocation anywhere in the world.

Frequently Asked Questions

Who is Ankur Sharma and what does he do?

Ankur Sharma is an Agentic AI & ML Engineer with a PhD from NTU Singapore and 8+ years of experience. He builds production multi-agent AI systems and reasoning-traceable AI for genomics and biomedicine. He is based in Singapore and open to relevant roles globally across industry, research, and startups. Contact: ankurs103@gmail.com or linkedin.com/in/ankurit.

What is sentinel-amr?

sentinel-amr is an explainable, agentic antimicrobial resistance classifier for bacterial pathogen surveillance. It predicts resistance from curated gene content using XGBoost, attributes every call to specific resistance genes via SHAP TreeExplainer, detects novel or out-of-distribution organisms with an alignment-free k-NN gate, and routes each sample through a bounded 5-node LangGraph agent that produces an auditable report. Trained on 298 real K. pneumoniae genomes from BV-BRC with ROC-AUC 0.894 on a group-aware holdout. Available at github.com/ankurgenomics/sentinel-amr under the MIT license.

What is agentic-genomics?

agentic-genomics (GenomicsCopilot) is an open-source LangGraph agent for reasoning-traceable variant interpretation. It takes VCF files and HPO phenotype terms and returns ranked, explainable reports of candidate genetic variants. It uses 7 deterministic nodes including ACMG-lite classification, Phrank HPO scoring, and a critic LLM for fact-checking. Available at github.com/ankurgenomics/agentic-genomics under the MIT license.

What is genomics-skills?

genomics-skills is a library of 8 agent-callable Python genomics tools: TCGA pan-cancer expression analysis (9,479 real patient samples), Kaplan-Meier survival analysis, GO/KEGG enrichment, PubMed search, protein variant mapping, 3D structure viewing, and volcano plots. Each skill has a SKILL.md contract and CLI entrypoint. LLM routing via Claude Haiku. Available at github.com/ankurgenomics/genomics-skills.

What is genome-ft?

genome-ft is a full weight-level fine-tuning of Nucleotide Transformer v2 (50M), a genomic foundation model from InstaDeep, on two GenomicBenchmarks DNA classification tasks. All 53.8M parameters are updated rather than using LoRA or a frozen head. The project emphasises a trustworthy evaluation protocol: a leakage-free train/validation/test split, the test set scored exactly once, three random seeds with reported variance, and the base model measured under the identical pipeline. It reaches 0.735 accuracy on enhancers (MCC 0.478 ± 0.003 over 3 seeds) and 0.872 on promoters (MCC 0.747). Code is at github.com/ankurgenomics/genome-ft and the model card is at huggingface.co/ankur0050/nucleotide-transformer-v2-50m-genomicbenchmarks-ft.

What technologies does Ankur Sharma work with?

Core technologies: LangGraph, Claude/Anthropic API, multi-agent orchestration, RAG, Python, PyTorch, scikit-learn, AWS (Batch, Lambda, S3, Step Functions), Nextflow, Docker, GitHub Actions. Domain expertise: clinical genomics (WGS/WES/RNA-seq/ChIP-seq), DRAGEN, GATK, variant interpretation, ACMG classification, regulatory validation (DVT, FDA compliance, HIPAA).

Is Ankur Sharma available for hire?

Yes. Ankur Sharma is open to relevant roles globally — across industry, research, and startups — where agentic AI, ML, or computational biology intersects with real-world impact. Based in Singapore (PR), open to remote, hybrid, or relocation anywhere in the world. Contact: ankurs103@gmail.com or linkedin.com/in/ankurit.

What results has Ankur Sharma achieved in production?

Key results: built a 5-agent AI system (GenomicsOps AI) as a personal project that reduced pipeline failure resolution time by 67% (from 3 days to 2 hours). Architected cloud genomics pipelines processing 6,000+ samples with 40% lower compute costs, 50% less storage, and 400 TB of data managed. Led clinical validation for oncology and prenatal diagnostic products with FDA compliance.