neuralll
Joined 5 years ago
Joined 5 years ago
109$ via cheapest TaxAct to send zero revenue or loss to IRS or face 25000$ fine for not filling 5472 is absurd
IRS search shown 6 agents.
Do they provide free e-filling for all taxpayers with <$72k income? = startups or small business ?
freetaxact.com no freefile.intuit.com no freetaxusa.com/freefile2020 no 1040now.net/freefile.htm no taxslayer.com/americanpledge no freetaxes.free1040taxreturn.com no
I filled IRS feedback to form 1120 got no response Apr 3, 2021, 5:54 PM got just Robotic Non-Response 1 from [email protected]: Apr 3, 2021, 6:30 PM Robotic Non-Response 1 from [email protected]
After lengthy email exchange with TaxAct I got sentences "Also, the IRS FREE FILE is for individual taxpayers, not businesses." or "if you use the IRS Free File | Internal Revenue Service product look up tool, you will see there is no category for business taxes" to which I replied If IRS or Free Filing Alliance actually meant that only individual taxpayers are eligible for free federal tax filing. They would simply clearly said so by using word “individual taxpayer” everywhere they used word “taxpayer” in sentences defining eligibility for free filing."
The best would be calling IRS and Free Filing Alliance by phone due to responding to emails.But mine engliish. sux.
Will some brave soul rise to the occasion?
As for the job-seeking part: I won't deny that I'm looking for opportunities. I'm currently an unemployed AI developer and could certainly use a job offer. But that wasn't the main reason for posting.
A few years ago I sold my house(which made me homeless) and moved to Australia after being told that completing an AI Master's degree there would provide a pathway to a post-graduate visa and eventually citizenship. After spending roughly $40k on tuition and living costs, the eligibility rules changed and the maximum age for that visa category was reduced from 48 to 35, leaving many international students without the pathway they had planned for. So yes, I'm job hunting, but this post wasn't intended as a recruitment ad.
The real motivation was technical excitement. Seeing models like GLM-5.3, K2, MiMo-2.6, DeepSeek and similar large MoE models approaching frontier-level capabilities while running locally at nearly 28 t/s on just 2×3090s plus DDR4 RAM felt genuinely liberating after years of dealing with cloud costs and usage limits.
I suspect many people with two or more GPUs will feel the same way. If we can make multi-GPU systems contribute compute instead of mostly acting as extra VRAM, a lot of users suddenly gain access to much faster local inference on models that previously felt impractical. That's really why I wanted to share it.