About
I build the platforms that let scientists focus on curing disease instead of fighting infrastructure. As a Principal Engineer with 12 years in enterprise cloud, I specialize in multi-agent AI systems and production-grade platform engineering for pharmaceutical R&D on AWS.
My current focus is shipping live multi-agent systems — purpose-built subagent roles, cross-session agent memory, and LLM observability — into drug discovery and digital pathology workflows. I also track the real cost of running AI at scale: FinOps tooling that maps token spend across model tiers, accounts, and teams. I've architected bioinformatics and MLOps platforms on AWS EKS, migrated critical research pipelines to production-grade Argo Workflows, and built multi-agent systems using Amazon Bedrock, Strands, and Claude.
Experience
Principal Engineer
Dec 2025 – Present
Johnson & Johnson Innovative Medicine · Beerse, Belgium
- Architected and shipped a live MVP multi-agent scientific research platform on Amazon Bedrock and AWS EKS, with purpose-built subagent roles (evidence gathering, analysis, synthesis, and an adversarial critic), cross-session memory via Amazon Bedrock AgentCore, LLM tracing in Langfuse, and role-based access control through Entra ID.
- Built an AI FinOps dashboard tracking Bedrock, Claude Code, and Kiro spend across seven AWS accounts — with automated multi-cloud cost classification and usage anomaly flags; recovered three days of silently lost cost data by diagnosing and fixing a root-cause defect in the Bedrock data extraction pipeline.
- Deployed a GPU digital-pathology AI inference pipeline on EKS using KEDA ScaledJobs for event-driven scaling and Karpenter for dynamic GPU node provisioning, generating interactive whole-slide image overlays and structured result data delivered back to the digital-pathology imaging platform.
- Stood up Kargo on EKS with Entra ID SSO and ArgoCD integration, and designed a dev-to-prod promotion pipeline for research workloads with automatic dev promotion and a manual approval gate for production.
- Built an internal LLM evaluation harness with domain-specific challenges across pharmaceutical R&D scenarios (regulatory, pharmacovigilance, bioinformatics, platform engineering), producing objective quality and cost scores to guide model selection.
- Delivered a portfolio-modelling AI agent (FastAPI + SvelteKit 5, Strands SDK) enabling analysts to query Monte Carlo simulation results in natural language, with anti-hallucination guardrails.
- Delivered the internal talk "When one AI isn't enough" on multi-agent collaboration, with a 17-slide deck, animated demo video, and AI-narrated walkthrough illustrating how specialised subagent roles outperform a single LLM on complex research queries.
- Driving adoption of AI-assisted engineering using Claude Code and Kiro across platform and data teams, accelerating internal tooling delivery and embedding agentic development patterns into the R&D engineering culture.
Solutions Architect (Partner)
Dec 2024 – Dec 2025
Amazon Web Services (AWS) · Brussels, Belgium
- Owned technical engagement and delivery success on partner implementation projects, developing deep expertise across AWS technologies and how applications are constructed on the platform.
- Defined implementation architectures and responded to RFPs, applying solution-design best practices and cloud-native use cases.
- Acted as strategic technical lead in the field, providing architectural guidance to accelerate partner engagements worldwide.
- Led deep architectural design reviews to ensure world-class, cloud-native solutions ready for production deployment.
- Managed customer and partner relationships through design, build, and migration to the AWS platform, driving strategic partnerships and cross-team collaboration.
Johnson & Johnson Innovative Medicine
Sep 2020 – Dec 2024 · 4 yrs 4 mos
Beerse, Belgium
Principal Engineer
Nov 2023 – Dec 2024
- Led the setup of AWS platforms used by scientists and biologists for data analytics and modeling, ensuring efficient and secure data processing.
- Implemented cost optimization strategies across 12 AWS accounts, delivering significant annual savings.
- Established Well-Architected Framework reviews across multiple projects, embedding security and architecture best practices.
- Partnered with AWS on testing and contributing to the Karpenter open-source autoscaler, substantially reducing EKS compute costs.
- Built a proteomics data-analysis pipeline on AWS EKS for the Bioinformatics team.
- Developed a platform for running Shiny applications (R, Python, Streamlit) on AWS EKS using open-source tooling.
- Led an ETL migration of HPC data from MongoDB to RDS PostgreSQL, powering PowerBI reporting and analytics.
Lead Engineer
Oct 2021 – Oct 2023
- Led setup of AWS platforms used by scientists and biologists for data analytics and modeling.
- Delivered cost optimization across multiple AWS accounts, generating significant annual savings.
- Established well-architected frameworks across projects to embed security and architecture best practices.
- Piloted new open-source technologies in collaboration with central platform teams.
- Implemented AWS DataSync to move data from on-prem NAS and instrument PCs into S3.
- Deployed Karpenter on EKS workloads for faster autoscaling and further cost optimization.
- Selected for the AWS Subject Matter Expert (SME) program to develop exam items for the AWS Solutions Architect Professional and AWS DevOps Engineer Professional certifications.
- Led an ETL migration of HPC data out of MongoDB into RDS PostgreSQL with PowerBI dashboards, informing workload analysis and helping secure project funding.
Senior IT Analyst
Sep 2020 – Oct 2021
- Containerized applications on AWS EKS, ECS, and Kubernetes across EC2 and on-prem instances.
- Deployed R and Python Shiny applications on Kubernetes via ShinyProxy.
- Implemented Kustomize and Helm charts for containerized workloads.
- Built GitOps workflows using ArgoCD, Argo Rollouts, and Argo Events.
- Designed cost-effective, performance-optimized, and secure solutions, including end-to-end encryption at rest and in transit and application-level authentication/authorization.
- Implemented observability with Prometheus and Grafana.
LTI — Larsen & Toubro Infotech
Oct 2017 – Sep 2020 · 3 yrs
Navi Mumbai, India → Beerse, Belgium
Senior Software Engineer — Beerse, Belgium
Jan 2019 – Sep 2020
- Worked as AWS Technical Specialist embedded at J&J (Belgium) for the EMEA region, with expertise in Redshift, EC2, S3, ETL (Talend), and Athena.
- Applied solution design and BI subject-matter expertise on the AWS platform, improving architecture and reducing total cost of ownership.
- Led migration of the entire AWS production environment to a new account, including Redshift, S3, EC2, RDS, EMR, and SWF domains.
Senior Software Engineer — Navi Mumbai, India
Oct 2017 – Jan 2019
- Worked as AWS and ETL developer, improving AWS architecture and reducing total cost of ownership.
- Identified opportunities to introduce AWS services for better platform performance.
Business Intelligence Developer
May 2017 – Oct 2017
Amdocs · Pune, India
- Built data-virtualization views across Teradata, Oracle, Redshift, Vertica, HDFS, and flat files using Cisco Composite for a multi-source BI project.
Cloud Engineer
Dec 2016 – May 2017
BlazeClan Technologies · Pune, India
- Delivered AWS cloud architecture and data profiling for a petrochemical operational-dashboard project using Talend, Redshift, and Tableau.
Capgemini
Apr 2014 – Dec 2016 · 2 yrs 9 mos
Pune, India
Senior Software Engineer
Apr 2015 – Dec 2016
- Built and maintained ETL pipelines (Datastage, Teradata BTEQ/MLOAD/FLOAD) and resolved production issues via root-cause impact analysis.
- Automated a manual job, cutting runtime from 45 hours to 30 minutes with improved accuracy.
Software Engineer
Apr 2014 – Apr 2015
- Early-career ETL development role.
Selected Writing