Tushar Soni

Software Engineering

[Andrew Ng @AndrewYNg]

Announcing OpenWorker! An open-source agent that doesn't just chat with you, but delivers finished work -- like hand you a polished document, send a slack message, or update a calendar entry.

Ask it to prepare a customer brief, untangle your calendar, draft a report, or triage a Slack alert. It works across your files and everyday tools, produces the deliverable, and checks in before doing anything consequential.

OpenWorker runs on your Mac, with Windows support coming soon. It does not lock you into any one model. Bring your own API key and run it with GPT 5.6 Sol, Claude Fable, Gemini 3.6, an open weight model (like Kimi, GLM, DeepSeek, Inkling), or Ollama to keep your data local. Your data does not leave your machine except through an LLM provider and integrations that you choose.

@rohitcprasad and I are building OpenWorker because AI coworkers are an important way to get work done, and we want there to be an open, privacy-preserving, model-independent option. Check it out and let us know what you think!

Try it out: https://t.co/P0mGnI1o31 (requires your own API key)
Source code: https://t.co/NYCiTD6hSq

[Paul Graham @paulg]

The biggest opportunity for would-be startup founders is AI. But the most underpriced opportunity is probably non-AI ideas. So if you have a good non-AI idea, go for it, because everyone else is going to overlook it.

[Fernando @Franc0Fernand0]

If you're a software engineer who wants to upskill in system design, read these 14 articles (links below):

[Tushar Soni @hacktushar]

@Punk2027 Done

[Tushar Soni @hacktushar]

@Crypto0637 Done

[Tushar Soni @hacktushar]

@belfort_max Done

[Indian Air Force @IAF_MCC]

The Indian Air Force (IAF) has successfully executed its assigned tasks in Operation Sindoor, with precision and professionalism. Operations were conducted in a deliberate and discreet manner, aligned with National Objectives.

Since the Operations are still ongoing, a detailed briefing will be conducted in due course. The IAF urges all to refrain from speculation and dissemination of unverified information.

#OpSindoor
#IndianArmedForces
#IndianAirForce
#OperationalCapability

@DefenceMinIndia
@MEAIndia
@MIB_India
@SpokespersonMoD
@HQ_IDS_India
@adgpi
@IndiannavyMedia

[Indian Tech & Infra @IndianTechGuide]

🚨 All media channels, digital platforms and individuals are advised to refrain from live coverage or real-time reporting of defence operations and movement of security forces

- Ministry of Defence, GOI.

[Ministry of Information and Broadcasting @MIB_India]

Ministry of Information and Broadcasting issues advisory to all Media channels to refrain from showing live coverage of defence operations and movement of security forces in the interest of national security.

[Tushar Soni @hacktushar]

Excited to announce that I’ll be participating in @perplexity_ai AI Business Fellowship to further develop my skills within AI.

[Tushar Soni @hacktushar]

Earned my certificate in Federated Learning through the #30DaysOfFLCode challenge! 🚀

Big thanks to the @openminedorg organizers and it's amazing community for driving the future of AI forward. Let’s keep pushing boundaries together! 🌟

[Tushar Soni @hacktushar]

Future directions focus on solving these challenges to create scalable, secure, and ethical AI systems. The integration of FL and FM has the potential to transform AI across industries! 🌍🚀

[Tushar Soni @hacktushar]

Final Day 30 of #30DaysOfFLCode :

Read 'When Foundation Models Meet Federated Learning' https://t.co/dQjlLpQiQE deep dive into their synergy
🔹 FL -> FM: Boosts privacy, expands data, enables collaboration
🔹 FM -> FL: Strong pre-trained start, synthetic data, better performance

[Tushar Soni @hacktushar]

Challenges:
🔹 Managing memory, communication, and computation costs for large FMs.
🔹Addressing data privacy, security, and IP concerns.

[Tushar Soni @hacktushar]

Day 29 of #30DaysOfFLCode :

Joined @openminedorg webinar Introduction to Privacy-Preserving Machine Learning
- Overview of privacy-preserving machine learning techniques
- Example applications & case studies
- Best practices for protecting sensitive data during model training

[Tushar Soni @hacktushar]

Day 28 of #30DaysOfFLCode :

Watched Scaling Federated Learning: Meta AI's Mike Rabbat explains how FL trains models across billions of devices while keeping data private. Innovations in personalization, efficiency, and privacy unlock AI at scale.
https://t.co/vsRdUNqcqH

[Tushar Soni @hacktushar]

Day 27 of #30DaysOfFLCode :

Explored interesting blog from Google - Advances in private training for production on-device language models https://t.co/M5vjkwElk4
and started with paper https://t.co/OqM32JaBrE to understand Differentially Private FL for Gboard Language Models

[Tushar Soni @hacktushar]

Day 26 of #30DaysOfFLCode :

Explored a interesting blog post on Learning with Privacy at Scale https://t.co/V21SyAZRry

Local differentially private algorithms:
- Private Count Mean Sketch
- Private Hadamard Count Mean Sketch
- Private Sequence Fragment Puzzle