mavam
Joined 14 years ago
Karma
167
Submissions
101
Member since
Sep 17
http://github.com/mavam
View on Hacker NewsRecent submissions
Comment · 8 months ago
Why choose if you can have both? You can turn any MCP into an CLI with Pete's MCPorter: https://mcporter.dev.
Since I've just switched from buggy Claude Code to pi, I created an extension for it: https://github.com/mavam/pi-mcporter.
There are still a few OAuth quirks, but it works well.
Comment · 9 months ago
Tenzir | Software Engineering & Sales | Remote (EU Timezones) | Full-time
Tenzir is building an open-core security data pipeline for the AI era. We empower defenders to collect, shape, and route security telemetry at scale. Our stack relies on C++23/26 and Apache Arrow for the core engine, with a modern TypeScript platform layer.
We are hiring for 4 key technical roles to join our distributed team (HQ in Hamburg):
- Software Engineer (Core Engine): Systems programming with C++23/26. Work on our pipeline language, query execution, database integrations, and streaming data processing with Apache Arrow.
- Software Engineer (Product): Full-stack TypeScript/Python focus. Build the control plane, platform APIs, and developer tools, AI workflows, and anything else that makes our engine accessible.
- Account Executive: Our first dedicated AE to own the full enterprise sales cycle. You'll work with complex deals in security and data infrastructure.
- Solutions Engineer: Lead POCs and post-sales implementation. You'll be the hands-on engineer ensuring customers succeed with complex data pipelines.
We are a remote-first team across European timezones. If you like solving hard data problems, we'd love to hear from you.
Comment · 9 months ago
Not yet, but will do so soon at https://github.com/tenzir.
Comment · 9 months ago
What's the startup latency? How long do I have to wait until Claude is operational?
Comment · 9 months ago
For deploying Claude Code as agent, Cloudflare is also an interesting option.
I needed a way to run Claude marketplace agents via Discord. Problem: agents can execute code, hit APIs, touch the filesystem—the dangerous stuff. Can't do that in a Worker's 30s timeout.
Solution: Worker handles Discord protocol (signature verification, deferred response) and queues the task. Cloudflare Sandbox picks it up with a 15min timeout and runs claude --agent plugin:agent in an isolated container. Discord threads store history, so everything stays stateless. Hono for routing.
This was surprisingly little glue. And the Cloudflare MCP made it a breeze do debug (instead of headbanging against the dashboard). Still working on getting E2E latency down.
Comment · 2 years ago
Tenzir | Remote - EU | open-core | Full-time | https://tenzir.com
Tenzir is hiring several key engineering roles to meet the needs in expanding the team. Our product: security data pipelines. From the data side, think of it as an Arrow-native, multi-schema ETL tool that offers optional storage in Parquet/Feather. From the security perspective, think of it as a solution for collecting, parsing, transforming, aggregating, and routing data. We typically sit between the data sources (endpoint, network, cloud) and sinks (SIEM, data lake).
Our open-source execution engine is C++20, our platform is SvelteKit and TypeScript. Experience with data-first frontend apps is a great plus. Open positions at https://tenzir.jobs.personio.de:
- Senior C++ Engineer
- (SecOps) Solution Engineer
Comment · 2 years ago
We see this trend as well. And AWS Security Lake goes exactly there.
Right now, we‘re working on OCSF normalization in our pipelines to drop structured security telemetry in the right format where you need it. Like a security ETL layer.
We considered ClickHouse and DuckDB but struggled with making the execution engine multi-schema, e.g., more jq-like but still on top of data frames. So we started with a custom catalog and engine on top of Parquet and Feather that we will later factor into a plugin to transpile our query language (TQL) to SQL. The custom language because security people are not data engineers.
Comment · 2 years ago
We're building something similar at Tenzir, but more for operational security workloads. https://docs.tenzir.com
Differences to Vector:
- An agent has optional indexed storage, so you can store your data there and pick it up later. The storage is based on Apache Feather, Parquet's little brother.
- Pipelines operators both work with data frames (Arrow record batches) or chunks of bytes.
- Structured pipelines are multi-schema, i.e., a single pipeline can process streams of record batches with different schemas.
Comment · 2 years ago
Tenzir | Remote (EU) or Hamburg, Germany | open-core | Full-time | https://tenzir.com
Tenzir is hiring several key engineering roles to meet the needs in expanding the team. Our product: security data pipelines. From the data side, think of it as an Arrow-native, multi-schema ETL tool that offers optional storage in Parquet/Feather. From the security perspective, think of it as a solution for collecting, parsing, transforming, aggregating, and routing data. We typically sit between the data sources (endpoint, network, cloud) and sinks (SIEM, data lake).
Our open-source execution engine is C++20 (https://github.com/tenzir/tenzir), our platform is SvelteKit and TypeScript. Experience with data-first frontend apps is a great plus. Open positions at https://tenzir.jobs.personio.de:
- Fullstack Engineer
- Senior Frontend Engineer
- UI Designer
- Solution Engineer
Comment · 2 years ago
We're developing TQL (Tenzir Query Language, "tea-quel") that is very similar to PQL: https://docs.tenzir.com/pipelines
Also a pipeline language, PRQL-inspired, but differing in that (i) TQL supports multiple data types between operators, both unstructured blocks of bytes and structured data frames as Arrow record batches, (ii) TQL is multi-schema, i.e., a single pipeline can have different "tables", as if you're processing semi-structured JSON, and (iii) TQL has support for batch and stream processing, with a light-weight indexed storage layer on top of Parquet/Feather files for historical workloads and a streaming executor. We're in the middle of getting TQL v2 [@] out of the door with support for expressions and more advanced control flow, e.g., match-case statements. There's a blog post [#] about the core design of the engine as well.
While it's a general-purpose ETL tool, we're targeting primary operational security use case where people today use Splunk, Sentinel/ADX, Elastic, etc. So some operators are very security'ish, like Sigma, YARA, or Velociraptor.
Comparison:
users
| where eventTime > minus(now(), toIntervalDay(1))
| project user_id, user_email
export
where eventTime > now() - 1d
select user_id, user_email
[#] https://docs.tenzir.com/blog/five-design-principles-for-buil...
Comment · 2 years ago
Tenzir | Remote - EU | open-core | Full-time | https://tenzir.com
Tenzir is hiring several key engineering roles to meet the needs in expanding the team. Our product: security data pipelines. From the data side, think of it as an Arrow-native, multi-schema ETL tool that offers optional storage in Parquet/Feather. From the security perspective, think of it as a solution for collecting, parsing, transforming, aggregating, and routing data. We typically sit between the data sources (endpoint, network, cloud) and sinks (SIEM, data lake).
Our open-source execution engine is C++20, our platform is SvelteKit and TypeScript. Experience with data-first frontend apps is a great plus.
Apply at https://tenzir.jobs.personio.de for the following positions:
- Fullstack Engineer
- Senior Frontend Engineer
- UI/UX Designer
- SOC Solution Engineer
We are based out of Hamburg, Germany, and hire across EU time zones, stretching all the way to India.
Comment · 3 years ago
The extra data type of unstructured/bytes is optional in that you don’t have to use it if you don’t need it. Just start with a table if that’s your use case.
In security, binary artifacts are common, e.g., to scan YARA rules on malware samples and produce a structured report (“table”). Turning packet traces into structured logs is another example. Typically you have to switch between a lot of tools for that, which makes the process complex.
(The “void” type is only for symmetry in that every operator has an input and output type. The presence of void makes an operator a source or sink. A “closed” pipeline invariant is one with source and sink, and only closed pipelines can execute in our mental model.)
Comment · 3 years ago
We want to achieve something similar with our pipelines [1] by making the beginning and the end of the pipeline symmetric, giving you this flow:
1. Acquire bytes (void → unstructured)
2. Parse bytes to events (unstructured → structured)
3. Transform events (structured → structured)
4. Print events (structured → unstructured)
5. Send bytes (unstructured → void)
The "Publish" part is a combination of (4) and (5). Sometimes they are fused because not all APIs differentiate those steps. We're currently focusing on building blocks (engine, connectors, formats) as opposed to application-level integrations, so turnkey Reverse ETL is not near. But the main point is that the symmetry reduces cognitive effort for the user, because they worked that muscle on the "E" side already and now just need to find the dual in the docs.
[1] https://docs.tenzir.com/blog/five-design-principles-for-buil...
Comment · 3 years ago
Author here: it’s describing how we built our pipeline engine. Technical content, presented in a form that I think is of interest to many HN users. I don’t think „ad“ is the correct term to refer to this. Can you elaborate what prompted your judgement?