- Interested in distributed systems, cloud-native infrastructure, AI agents, and developer tools
- Exploring how AI systems and infrastructure can work together at scale
- I learn mostly by building things, breaking them, and figuring out why they broke
| OpenAlps |
Document-grounded claim verification over Swiss official voting booklets. Given a claim and its voting booklet, the system decides whether the booklet supports it, contradicts it, or says nothing either way. It works in four stages: extract and split booklet PDFs into sentence-level evidence with page references, retrieve evidence across German, French, and Italian by combining lexical matching with multilingual embeddings, classify the claim with Apertus v1.5 models, and validate the generated evidence against the source text before writing structured predictions. |
| PathFinder |
A search engine for human experiences, built on a GraphRAG-powered Life Graph. Turns real career and life journeys into a connected Neo4j knowledge graph of goals, experiences, skills, decisions, and proofs, so people can ask what actually happened to others in their situation. Voice or chat input is transcribed with Sarvam AI and structured by Gemini, proofs such as GitHub, PDFs, and images are verified before they enter the graph, and hybrid GraphRAG retrieval answers queries with relevant journeys and decision patterns. Ships as a cross-platform Expo app for web, iOS, and Android. |
| K8Lab |
A visual Kubernetes lab for inspecting cluster state and reproducing failure scenarios. Brings resource inspection, topology visualization, logs, and activity history into one interface over the Kubernetes API. Explore Node, Deployment, ReplicaSet, Pod, and Container relationships as an interactive graph, manage resources and multiple clusters, and run reproducible experiments such as CrashLoopBackOff, ImagePullBackOff, OOMKilled, Pending Pod, and failed liveness probes, with live logs streamed from the cluster. |
| Krkn |
Open source contributor to a CNCF chaos engineering framework. Contributed across the krkn, krkn-ai, and website repositories: chaos scenario fixes, rollback reliability improvements, and documentation. Fixed rollback and PVC scenario bugs, including adding fallback pod selection to the PVC scenario. |
| Languages |
|
| Web & Backend |
|
| Databases |
|
| Cloud & Infra |
|
| AI / Agentic |
|
