Arindam Dey

Azure, GCP, Python, .NET, React,
Microservices, Docker, Kubernetes, Redis, Cosmos
Scalable SaaS volunteer management platform for onboarding, engagement, scheduling and
analytics, on microservices across multi-cloud.
• Led solution architecture for a multi-cloud (Azure, GCP) microservices SaaS platform
serving 50, 000+ users at 99.9% uptime, with production Generative AI for engagement and
insight.
• Architected a domain-agnostic AI matching engine - a single reusable platform that
matches any entity type (volunteer-opportunity, job-resume, mentor-mentee, patient
hospital) via vector embeddings and semantic search with an LLM explanation layer (Azure
OpenAI GPT-4o), cutting manual matching effort ~60%.
• Designed RAG and agentic workflows (LangChain, LangGraph, Semantic Kernel) with
prompt engineering and human-in-the-loop validation, automating ~70% of routine
decisions while keeping outputs auditable.
• Implemented LLM observability and per-request cost governance and applied parameter
efficient fine-tuning (LoRA, QLoRA), reducing inference cost ~35% and improving matching
accuracy ~20%.
• Built a single configuration-driven (JSON) generic web scraper that adapts to any new
source or use case without code changes - competitor advertising, rival news, financial
sentiment - with NLP and LLM sentiment analysis across 100+ sources, replacing ~20
hours/week of manual research.
• Designed a centralised volunteer-management SaaS that integrates any authentication,
CRM or data source, deployable as a single-tenant enterprise or multi-tenant SaaS product.
• Owned CI/CD and containerization (Docker, Kubernetes); grew and mentored a team of 8,
setting the technical roadmap and architecture standards.

Kubernetes, MongoDB, Synapse, Kinesis
Criminal-record check and data-analytics platform for a government client.
• Directed account-level solution and integration architecture for a criminal-records and data
analytics platform processing 1M+ background checks, with hands-on delivery.
• Architected an automated .NET migration tool exceeding Microsofts Upgrade Assistant in
code analysis and presales estimation; adopted across business units, it supported 271
projects and generated $10M in its first quarter.
• Partnered with Microsoft to migrate a legacy, unsupported BizTalk estate to cloud via a 3-
year support agreement that spread cost across three financial years and boosted client
fiscal-year profit.
• Built APIs to calculate carbon emissions (devices, electricity, employee travel) for U.S.
Green IT carbon-reporting compliance; drove presales and authored .NET Practice best
practices across presales, development, architecture and infrastructure.

Renewable Energy. .NET Core 3.1, React, Azure, Python, Data Science, Data Lake Gen2, Data Factory
Imputed missing historical weather-forecast data via predictive modeling to drive correct turbine-model selection and design.
• Built data-science capability to impute 10+ years of missing weather data via predictive modeling at ~92% accuracy, improving turbine-model selection; served as cloud and integration architect.
• Migrated on-premises workloads to Azure: release cycles cut from 1 week to 2 days, 99.99% uptime, deployment failures down from 18% to 2%, MTTR down 60%, 25+ recurring incidents eliminated in 3 months; championed SAFe adoption and mentored 9.
• Designed a robust Azure data interface for seamless exchange between legacy systems and modern applications, ensuring business continuity and accelerating digital transformation.

RabbitMQ, Redis, React,
Redesigned a monolith into a microservices, cloud-resilient system for vendor sales.
• Re-architected a monolith into microservices, cutting downtime 70% and scaling to 3x peak
traffic; solution architect, Agile coach and Scrum Master for a team of 10.
• Delivered account-level Agile training (Scrum with Kanban and SAFe) and implemented
structured client- and server-side logging (Serilog, JSNLog).

Migration of a retail platform into AWS cloud.
• Delivered an AWS cloud migration as project manager, application and integration
architect, running enterprise SAFe rollout and a 7-person team for a newly grown account.

Sales.
myConcerto sales-automation and revenue-forecasting platform modeling business and revenue
across onboarded and prospective Fortune 500 clients.
• Served as cloud architect and full-stack lead for myConcerto, an internal sales-automation
and revenue-forecasting platform used across 100+ Fortune 500 engagements that
generated roughly one-third of global Accenture revenue.
• Drove enterprise SAFe adoption as Agile coach across an 800+ member account and
managed a 10-person team; recognized with the ACE Award for zero defects and
escalations.

Retail.
JavaScript, C++, COM
Mobile application across all chain sites of a national drugstore retailer.
• Delivered a multi-site retail/drugstore mobile application as service-oriented architect,
owning design, architecture and delivery documentation; Scrum Master for a team of 9.
• Acted as SME for existing-tool enhancement and application-architecture team member;
trained and onboarded new joiners and maintained delivery documentation in the absence
of an application wiki.