Senior engineer, 13+ years in distributed systems. I build anti-financial-crime systems at a Nordic bank by day, and solo-ship agentic-AI products — and win hackathons — by night.
I care about AI that explains its reasoning, leaves an audit trail, and actually works in production.
- 🏦 Software Engineer @ Swedbank — anti-financial crime & AML
- 🤖 Building agentic AI on Azure AI Foundry, A2A, and MCP
- 🧮 Researching GPU & quantum compute — q1729, the quantum taxicab
🛡️ ARGUSMulti-agent compliance intelligence on Azure AI Foundry Microsoft Agents League — AI Skills Fest 2026 · Reasoning Agents track ↗
problem> Manual KYC/AML review doesn't scale, and unaudited AI decisions don't survive a regulator's questions. approach> Five specialist AI agents, coordinated over the A2A protocol on Azure AI Foundry, run KYC, AML, and risk screening — every finding grounded in cited regulation via Foundry IQ, every decision backed by a full audit trail. Built solo; one of three Hack for Good winners. Studied Foundry IQ before building ARGUS, then competed in the Agents League Reasoning Agents track — and won Hack for Good (1 of 3). Foundry IQ Learned before building ↗ · Reasoning Agents Track participant ↗
|
🔎 DRIFTRelease intelligence for GPU & AI infrastructure problem> GPU/AI-infra changelogs are noisy and scattered — teams miss the one release that breaks their pipeline. approach> An agentic pipeline that turns noisy changelogs into cited, bounded, inspectable release briefings for GPU and AI-infrastructure teams — grounded and source-linked, never hand-wavy.
Repository ↗ · Watch demo ↗ · Live app ↗ · API docs ↗ · Devpost ↗ |
|
AI-powered financial storytelling Backblaze Generative Media · Devpost ↗
problem> Nobody reads a transaction CSV for fun — but everyone will watch a 60-second recap of their own year. approach> Spotify Wrapped, but for your bank account — an agentic pipeline that turns a transaction CSV into a personalized 60-second narrated recap video. Three of four AI steps route through the Genblaze SDK; Backblaze B2 is the source of truth, storing 14 artifacts per session with a SHA-256 apiece, so share links survive a full backend redeploy.
Repository ↗ · Watch demo ↗ · Live app ↗ · API docs ↗ · Devpost ↗ |
Durable incident memory for cold-started agents problem> An agent killed mid-incident normally loses all context — and starts the investigation over from zero. approach> Agentic incident-response memory that survives the agent being killed mid-incident — execution state lives in CockroachDB, not process memory, so a cold-started agent resumes from the exact interrupted step.
|
-FFB000?style=flat-square){kind=icon}
-
ARGUS: Compliance Infrastructure That Believes Financial Access Is a Human Right — techcommunity.microsoft.com · Guest post (July 2026)
Microsoft published my full write-up on the Educator Developer Blog, including how ARGUS coordinates five agents over A2A with citation-grounded risk scoring.
Lee Stott · Microsoft in #agentsleague (theme-aware image)
|
Ramanujan's mathematics meets the NVIDIA stack
problem> How fast can a 100-year-old series actually run on today's silicon — classical and quantum? approach> How fast can a GPU compute π — classically, and as a quantum computer? An end-to-end study from consumer RTX to datacenter H100 (CUDA C++ and CUDA-Q/cuQuantum), with an AI layer that writes up what the numbers actually show.
|
🧩 Agentic AI & Enterprise Intelligence I design AI systems with Azure AI Foundry, Foundry IQ, multi-agent orchestration, Agent-to-Agent (A2A) communication, and RAG with hybrid search — built to be explainable, grounded, and production-ready.
🏗️ Distributed Systems & Backend Architecture 13+ years building scalable platforms with Java (Spring Boot, Quarkus) and Python (FastAPI) — microservices, NoSQL, event-driven systems, and hybrid cloud across AWS, Azure, and OpenShift.
⚙️ AI Infrastructure, Performance & Compute I work at the infrastructure layer behind modern AI — GPU computing, NVIDIA CUDA, model serving, vector search, and performance engineering.
📡 Reliability, Observability & Platform Engineering I build resilient systems with Azure Functions, Service Bus, OpenTelemetry, Application Insights, and KQL — telemetry pipelines that hold up to enterprise-grade reliability and governance.
Languages & Frameworks
Agentic AI & LLM
Cloud & Infrastructure
AI Infra, GPU & Observability
2025 – Present · Software Engineer · Swedbank — Stockholm, Sweden Anti-financial crime · AML platforms · 95%+ test coverage across a unified multi-module architecture
2021 – 2025 · Senior Java Developer · Viaplay Group — Stockholm, Sweden Media & streaming platforms on AWS + Kubernetes · ~30% performance and ~40% delivery-speed gains
Mar–Sep 2021 · Software Developer · Expleo Technology Nordic — Gothenburg, Sweden Domain-driven microservices · ~50% faster onboarding via documentation & workflow diagrams
2012 – 2021 · Senior Software Engineer · 8 yrs 6 mos · IBM — Sydney & Pune Regulated banking APIs for Westpac · ~25% response-time gains · Jenkins/Bamboo delivery modernization
- q1729 — Ramanujan × the NVIDIA stack — CUDA-Q/cuQuantum simulation with an AI analysis layer
- llm-qlab — LLM quantization benchmarks on consumer GPUs — speed, VRAM, and quality trade-offs
- pythonic-algorithms-lab — CPU vs GPU algorithm benchmarking with empirical Big-O profiling (CuPy + Numba CUDA)
- iq-series — Hands-on Microsoft IQ notebooks — Foundry IQ, Work IQ, Fabric IQ
- NVIDIA Certified Professional: Agentic AI (NCP-AAI) — NVIDIA
- Machine Learning and AI Foundations: Prediction, Causation, and Statistical Inference — LinkedIn Learning Community
- IBM Machine Learning Essentials — IBM
- Python for Data Science — IBM
- Kubernetes Administration (LFS458) — The Linux Foundation
Full certification list on LinkedIn ↗
Building trustworthy AI systems that explain their reasoning, leave an audit trail, and actually work in production.
If that's the kind of problem you're working on — I'd love to talk.
github.com/iarjunganesh · arjunganesh.dev