Two tracks in parallel across my career: building software and shaping policy. Started in embedded systems and DevOps (CI/CD, cloud-native infra, security auditing), crossed into government - advised the Mayor of Bucharest on a city-wide digital parking overhaul and served on the executive board of the municipal parking company.
Now back to building: working part-time at an EdTech startup applying the modern AI stack (Python, TypeScript, LLM APIs, RAG, agentic workflows) to real products on GCP.
Engineering depth + public-sector leadership + hands-on AI. Looking for a role where technical ability and strategic thinking both matter - ideally at the intersection of AI and real-world operations.
MSc European Politics (London School of Economics), Engineer's Diploma in Systems & Computers (Politehnica Bucharest). English C1-C2 (TOEFL 112/120).
From a business strategy perspective, it makes a lot of sense not to eat the credit card risk, otherwise you commit yourself to a very bad position: scammers will signup on your platform as both buyers and sellers and effectively launder stolen credit cards.
From a labor rights perspective, they are clearly not a small business that can take contractual risk, they are a de facto employee dependent on the revenue the Upwork platform generates. Withholding payment for work done is a major abuse.
Each side has their truth and I cannot see how it can be resolved in the general case without regulatory intervention in the area of platform labor.
There is no fundamental reason for urban sprawl to stop. Western living standards and wealth are increasing, and the actual material costs of laying a sewer pipe or road become negligible.
With the advent of electric self driving cars, transportation becomes cheap and hassle free. A network of highways converging to an urban center served by Boring co Loop tunnels can gather trafic from a 30 mile area. As long as travel times are less than say, around 45 minutes, those people will be living "in the city".
A 30 mile radius means about 7 million plots of 10000 sq feet, so you could accommodate more than 20 million inhabitants in this suburbopolis of single detached family homes.
The factors preventing it are land prices, regulatory pressure and homebuyer tastes.
If you can automate 80% of the jobs your customers require, you can make a billion dollars and send the other 20% potential customers to talk to Mailer Daemon. If you automate 80% of every job and leave 20% to every customer, you have no customers.
The difference between sync level and black level in analog TV standards is about 15% of maximum transmitter power. This means that any noise added that is more than 7.5% of max power will confuse the TV to detect sync as black and black as sync. In practice, a 20dB SNR or less will be completely unwatchable.
The Shannon limit for a 5Mhz channel with 20 dB SNR is 17 Mbit/s and latest DTV standards approach this, easily 10, maybe 15 Mbit. That's enough for a HD channel in perfect quality or many more in SD.
So the problem is not with analog vs digital, that debate is long settled. It's the specific tradeoffs made when digital TV was deployed.
They are built upon the same basic principles - QAM modulation, error correcting codes and MPEG/HEVC codecs, so there should be no significant difference.
The newest ATSC3.0 standard has a somewhat richer toolbox allowing the optional use of 1024 or 4096QAM vs 256QAM for DVB, more advanced LDPC error correcting codes, LDM multiplexing and MIMO. The industry seems to believe the added complexity is not worth the marginal benefits.
The fundamental difference between Europe (or at least UK) and US seems to be the concern regulators had for improving over the air TV. So while in the UK, the move to digital was employed to broadly improve the offering and reception quality, in the US there was less concern and the stations found the death of analog a good opportunity to close down transmitters, recover spectrum and remove free programing.
There are many factors that come into this. The bandwidth of a single analog channel can accommodate 4 or more digital channels. If the transmitter is not upgraded, its power will now be spread to multiple channels.
Another is simply self-selection bias. When a rapid technical transition takes place, the average reception might improve significantly, but you can't guarantee it for everyone. A slight frequency change, a reflection pattern can drastically alter the signal intensity in a certain spot. But you will only hear from people who happened to have good analog and lost it, those with bad analog had already moved to cable and are unaware that good terrestrial digital has become available.
My experience with Freeview in the greater London area was very positive, with a very simple antenna I was able to get around 100 channels, the vast majority in perfect quality.
Snowy analog with intelligible sound translates to about 40dB signal to noise ratio. That's a SNR where digital will deliver crystal clear image. When digital becomes glitchy, you will absolutely be unable to watch analogue, you will have trouble maintaining vertical and horizontal sync and the sound will be drowned in noise.
Remote: Yes
Willing to relocate: Open to discuss
Technologies: Python, TypeScript, PHP, SQL, Rust, Bash. LLM integration RAG pipelines, agentic workflows. GCP (Cloud Run, Cloud Functions, Firestore,Cloud Storage), AWS (Lambda, S3, EC2), Azure. Docker, CI/CD, infrastructure-as-code. MySql, PostgreSQL, pgvector, Redis, Elasticsearch. Security hardening, GDPR. Hardware, low level control, electronics, embedded, robotics, networking and sysadmin etc.
Résumé/CV: https://www.linkedin.com/in/sbene/
Email: [email protected]
Two tracks in parallel across my career: building software and shaping policy. Started in embedded systems and DevOps (CI/CD, cloud-native infra, security auditing), crossed into government - advised the Mayor of Bucharest on a city-wide digital parking overhaul and served on the executive board of the municipal parking company. Now back to building: working part-time at an EdTech startup applying the modern AI stack (Python, TypeScript, LLM APIs, RAG, agentic workflows) to real products on GCP.
Engineering depth + public-sector leadership + hands-on AI. Looking for a role where technical ability and strategic thinking both matter - ideally at the intersection of AI and real-world operations.
MSc European Politics (London School of Economics), Engineer's Diploma in Systems & Computers (Politehnica Bucharest). English C1-C2 (TOEFL 112/120).