01KYC request
025 specialists
03Cited report
ARGUS
Multi-agent compliance intelligence
Citation-grounded regulatory lookups with reproducible audit trails.
ProofHack for Good winner — 1 of 3
Python 3.11Azure AI FoundryFoundry IQAzure OpenAI GPT-4o
Stockholm · Engineering production systems since 2012
I build auditable AI and distributed systems.
Anti-financial-crime engineer at Swedbank, independently building governed agentic systems and AI infrastructure that remain explainable under real constraints.
14+Years shipping
1 of 3Hack for Good winners
Swedbank · Viaplay · IBMBanking, streaming, and distributed platforms
Selected work
All systems Three systems. Three explicit trust boundaries.
01IAM event
02Durable ID
03Human review
BASTION
A governed institutional-agent fleet for continuous access review
Human review receives minimized risk categories rather than raw IAM bindings.
Python 3.12Google ADK 2.8Gemini 3.5 FlashVertex AI
01Incident alert
02Durable step
03Cold resume
CONTINUUM
Durable incident memory for cold-started agents
Resumes an interrupted incident from the exact step it was killed on.
Python 3.14FastAPICockroachDBAWS Lambda
External validation
Microsoft Agents League · Hack for Good
ARGUS was selected as one of three Hack for Good winners, then featured through a Microsoft Tech Community guest post on explainable compliance infrastructure.
Read the recognition →How I work
All principles The boundaries come before the model.
Deterministic core, probabilistic edge
Use models to explain, classify, summarize, or route. Keep authorization, scoring, and state transitions explicit where risk demands it.
Auditability by construction
Treat evidence, citations, and decisions as first-class records rather than trying to reconstruct them from logs later.
Durable state over ephemeral memory
Design agents and workflows to survive retries, cold starts, and asynchronous boundaries without losing investigation identity.
Career
14+ years across systems that cannot hand-wave failure.
Regulated banking at Swedbank and IBM. Streaming infrastructure at Viaplay. Independent investigations across governed AI, durable workflows, and GPU compute.
View the career timeline →Personal note
Time makes the constraint real.
Bhagavad-gītā 11.32 is the reminder behind this portfolio: systems, decisions, and reputations all meet time. Build what can still be explained when they do.
Bhagavad-gītā · Chapter 11, Verse 32
श्रीभगवानुवाच ।कालोऽस्मि लोकक्षयकृत्प्रवृद्धोलोकान्समाहर्तुमिह प्रवृत्तः ।ऋतेऽपि त्वां न भविष्यन्ति सर्वेयेऽवस्थिताः प्रत्यनीकेषु योधाः ॥
śrī-bhagavān uvācakālo 'smi loka-kṣaya-kṛt pravṛddholokān samāhartum iha pravṛttaḥṛte 'pi tvāṁ na bhaviṣyanti sarveye 'vasthitāḥ pratyanīkeṣu yodhāḥ
The Supreme Personality of Godhead said: Time I am, the great destroyer of the worlds, and I have come here to destroy all people. With the exception of you [the Pāṇḍavas], all the soldiers here on both sides will be slain.Bhagavad-gītā As It Is 11.32
Bhagavad-gītā As It Is, A.C. Bhaktivedanta Swami Prabhupāda — © The Bhaktivedanta Book Trust
Let’s build something auditable.
Open to senior engineering opportunities, governed-AI collaboration, and technical research conversations.